Skip to content

Track record

In every brief we write in advance what would disprove that reading, and in every report the condition that breaks the causal chain. When they come due we record the outcome here. Records where we were wrong are not deleted or left out of this page.

So far

546 checks · 98 records reviewed

Confirmed
63
The published reading survived the data that followed.
Partly borne out
157
Part of the reading held, part of it did not.
Falsified
80
The disproving condition occurred; the reading is void.
Still open
246
The horizon has not passed, or the decisive data is not in yet.

The sample is small. These figures are not a success rate, only a breakdown of the checks made so far.

Why we do this

A claim that cannot be disproved is not analysis. Writing down, at the moment of publication, what would invalidate a reading makes it impossible to bend the story to fit events afterwards. This page shows whether that promise was kept.

Calibration

Giving a probability is a different claim from being right. If half of the things we called "50%" happened, the probability was set correctly; if all of them happened, we were too cautious. The table below splits our published scenario calls by the probability we gave and shows what happened in each band.

Probability we gaveClosed callsRealisedConfirmedPartlyFalsifiedOpen
%0–2430%272121631
%25–3938%170132541
%40–5429%50225226
%55–10026%48025114

Only scenario calls that have come due are counted (123). Open calls do not affect the rate; they enter the table as they mature.

The realised rate counts "partly borne out" as a half. That is a judgement; the raw breakdown sits beside it so you can recompute with your own weighting.

Where the realised rate in the top band falls below the probability we gave, we are overconfident at high probabilities. The way to fix that is not to change the table but to give more cautious probabilities.

Every call

Every check we have published: the day it ran, what was claimed, the probability given and what happened when it came due. Falsified records are not deleted.

Download the table (CSV)

The full record, one file. Opens in any spreadsheet.

CSV ↓

Record seal · Record as of 11 November 2026

0475805d457a0b3ccb2bc77a69f3007c93decd28bc19ad782ed5c3b25314039d

This is a fingerprint (SHA-256) of today's complete record. Download the CSV and keep it: if the seal ever changes, past records have been altered. We generate the seal ourselves, so this is not an independent audit — but it does let you run the check without needing our permission.

PublishedTypeProbabilityVerdictClaim
11 November 2026Watched indicator—ConfirmedThe one thing to watch: the US 30-year auction at 21:00 on 12 November. An indirect bidder share below 70% would confirm that foreign demand is pulling back and that the term premium is widening independently of the Fed decision.
11 November 2026Falsifier—Partly borne outFalsifier: an extension of the US–China suspensions to 12 months at APEC in Shenzhen, together with an indirect bidder share back above 70% at the 12 November 30-year auction, would refute the 'easing on lease' reading.
11 November 2026Scenario%50Partly borne outH1 · 50% — Easing on lease continues: the suspensions are renewed for a short term, the Fed hikes by 25 basis points in December and Brent stays at 85–95 dollars.
11 November 2026Scenario%30Still openH2 · 30% — Terms lengthen and easing becomes durable: the suspensions are extended to 12 months, foreign demand returns at the 30-year auction and Hormuz traffic exceeds half of pre-war levels.
9 November 2026Scenario%55Still openH1 · 55% — Talks open, skies open: the Abu Dhabi round ends with a new schedule of talks but no ceasefire; strikes on the grid and bridges continue through the winter.
9 November 2026Scenario%30FalsifiedH2 · 30% — A narrow infrastructure ceasefire: the sides agree a limited ceasefire, similar to the Oleshky model, covering energy facilities and possibly ports.
9 November 2026Causal chain—Partly borne outSecond link: grid damage pushes Kyiv to prioritise a narrow energy and infrastructure ceasefire in Abu Dhabi and later rounds; Moscow tries to tie this to other concessions.
9 November 2026Causal chain—Partly borne outThird link: if an infrastructure ceasefire excludes ports, port strikes keep the insurance cost of Black Sea trade high; the indicators to watch are the war-risk premium and daily Bosphorus transits.
9 November 2026Watched indicator—Still openThreshold: January ICE Brent above 105 dollars would increase US pressure over strikes on Russian refineries and narrow Kyiv's logistical leverage.
9 November 2026Falsifier—Still openBreaking condition: if the mid-November Ukraine Defense Contact Group meeting yields new Patriot systems or interceptor deliveries, the interception rate rises and the chain breaks at the first step.
9 November 2026Watched indicator—Partly borne outPrimary indicator: if US October CPI at 16:30 on 10 November beats the 0.3% monthly forecast, a December hike is locked in and the term premium reopens at that evening's $42 billion 10-year auction.
9 November 2026Causal chain—FalsifiedThe fastest channel to react to a hot US CPI will be emerging-market currencies and Türkiye's external financing costs.
9 November 2026Watched indicator—Partly borne outSecondary indicator: whether the 10–11 November Abu Dhabi round produces an energy infrastructure ceasefire clause will determine whether the Black Sea war premium drops out of its 0.6–1% band.
9 November 2026Scenario%50Still openH1 · 50% — Prices ease, risk shifts to the public sector: Brent stays at $90–100 and Gulf political violence capacity is stuck at $200–400 million.
9 November 2026Scenario%30Still openH2 · 30% — The gap closes in price's favour: a Washington–Tehran framework reopens Hormuz, the tanker war premium falls below 2% and Brent drops below $90.
9 November 2026Scenario%20Still openH3 · 20% — The gap closes against price: an onshore Gulf facility is heavily damaged, or October CPI locks the Fed into a December hike and the dollar jumps; Brent tops $100 and CDS premiums widen.
9 November 2026Falsifier—Still openFalsifier for the combined signal: an increase in war and political violence retrocession capacity at the 1 January 2027 renewal, or Gulf political violence capacity returning above $500 million, would falsify the reading.
8 November 2026Scenario%50Still openH1 · 50% — Land crossings narrow, Iraqi volumes hold: Iraq closes several crossings, sanctions hit Caspian traffic and there is no major wave of attacks in Hormuz; Iraqi exports stay at 3–3.4 million barrels a day, but conditions harden.
8 November 2026Scenario%30Still openH2 · 30% — A Qatari formula loosens the blockade: the US accepts a step towards lifting the blockade, Iran grants IAEA access and opens Hormuz under an interim arrangement; benchmark oil falls below 90 dollars.
8 November 2026Scenario%20Still openH3 · 20% — Fighting resumes: the US restarts major operations before January and Iran retaliates against Gulf infrastructure; Basra loadings fall and benchmark oil rises above 110 dollars.
8 November 2026Watched indicator—Still openThresholds: January Brent above 110 dollars means a fresh Gulf outage is being priced, below 90 dollars an easing of the blockade; daily Hormuz transits below 20 mean shipowners are pulling out of Basra.
8 November 2026Causal chain—Partly borne outStep 1 (weeks): Baghdad closes more crossings or tightens controls; Iran's western border trade is squeezed into a handful of points and tension rises with Iran-aligned Iraqi groups.
8 November 2026Causal chain—Partly borne outStep 2 (weeks): Tehran tightens control in Hormuz, incidents in the strait multiply and the Gulf war risk premium holds in the 6–10% band or rises; buyers ask Iraq for delivery outside the strait.
8 November 2026Causal chain—Still openStep 3 (months): To protect revenue, Iraq leans on its own tankers and the Ceyhan line; Ceyhan loadings rise and Türkiye's transit income reflects the shift.
8 November 2026Falsifier—Still openBreaking condition: if Qatari mediation secures an easing of the blockade and an interim reopening of Hormuz, pressure on the land crossings loses its point and the chain breaks at the first step.
6 November 2026Scenario%50Still openH1 · 50% — Prices swing, buffers erode: Brent oscillates in a $90–105 band, the Gulf war premium stays at 6–10%, and the Yanbu outlet is threatened but stays open.
6 November 2026Scenario%30Still openH2 · 30% — A political framework replenishes buffers: the war premium falls below 2%, transits rise and Egypt recalls its deferred cargoes.
6 November 2026Scenario%20Still openH3 · 20% — Buffers break: the Yanbu outlet is hit or Türkiye's net reserves excluding swaps fall below $20 billion; Brent returns to its $113 peak.
6 November 2026Watched indicator—Still openPrimary indicator: if CBRT weekly data on 12 November show net reserves excluding swaps falling from $24.3 billion to below $20 billion, the currency anchor's capacity to carry the energy bill reaches its limit.
6 November 2026Falsifier—Still openConfluence falsifier: a rise in net reserves excluding swaps on 12 November, Egypt recalling its deferred cargoes and the Gulf war premium falling below 2% would disprove the reading.
5 November 2026Scenario%55Still openH1 · 55% — Prices fall, the strait stays shut: Iran does not narrow its blacklist, daily transits stay at 3–7 ships, the war-risk premium holds in the 6–10% band, and part of the drop to 94.62 dollars is recovered within weeks.
5 November 2026Scenario%30Still openH2 · 30% — Political framework, gradual reopening: the new balance in Washington speeds up framework talks with Iran, the blacklist narrows and transits with tracking signals on return to double digits.
5 November 2026Scenario%15Still openH3 · 15% — The list turns into seizures: Iran seizes one of the listed Gulf state tankers or stops vessels conducting ship-to-ship transfers; January Brent returns above 100 dollars.
5 November 2026Watched indicator—Still openPrimary indicator: if January Brent falls below 90 dollars while transits stay at 3–7 ships, the market is treating the constraint as overcome before we do.
5 November 2026Watched indicator—Still openSecondary indicator: the US affiliates rule and China's rare earth suspensions on 10 November; if no notice is published, access shifts from a contractual term to a formal restriction.
5 November 2026Falsifier—Still openFalsifier for the converging-signals reading: the reading weakens if the US and Chinese suspensions are extended by notice on 10 November, if the ex-China price of yttrium halves by the end of December, or if daily transits with tracking signals on return to double digits in Hormuz.
5 November 2026Scenario%50Still openH1 · 50% — The dark corridor persists: the blockade and the PGSA list stay in place, transits with tracking signals on stay below 10 a day, and Gulf exports hold in the 15–18 million barrels a day band.
5 November 2026Scenario%30Still openH2 · 30% — Deal and gradual reopening: the US and Iran reach an interim deal easing the blockade and sanctions, the PGSA list is suspended, and premiums fall with a lag as insurers wait for an incident-free period.
5 November 2026Scenario%20Still openH3 · 20% — The list cuts the transfer chain: the PGSA lists foreign tankers conducting transfers and seizes one, or an attack occurs; Gulf exports fall below 15 million barrels a day.
5 November 2026Watched indicator—Still openThreshold: if January Brent falls below 90 dollars while the blacklist grows, futures prices have decoupled from the delivery constraint.
5 November 2026Causal chain—Partly borne outStep one (days): listed state fleets turn to transits with tracking signals off and to ship-to-ship transfers at sea.
5 November 2026Causal chain—Still openStep two (weeks): the PGSA also lists foreign vessels taking over transfers, the war-risk premium does not fall below the 6–10% band, and the transfer link becomes more expensive.
4 November 2026Scenario%55Partly borne outH1 · 55% — The window closes and the blockade holds: Qatar's 36-hour window lapses unanswered, tanker attacks continue, the war premium stays at 6–10% and January Brent recovers part of its two-session fall.
4 November 2026Scenario%30FalsifiedH2 · 30% — Framework deal, delayed cover: the US and Iran reach a framework on the Qatari proposal and the strait reopens, but the Joint War Committee keeps its Gulf listing for weeks.
4 November 2026Scenario%15FalsifiedH3 · 15% — The window closes in conflict: attacks extend beyond Qeshm, or the Habshan–Fujairah line and Red Sea facilities are hit, and cover is withdrawn under 7-day cancellation clauses.
4 November 2026Watched indicator—FalsifiedPrimary indicator: if January Brent falls below $100 while the window lapses unanswered, the market has moved past the constraint before us; Brent at $103.60 does not translate into physical supply until the war premium drops below 2%.
4 November 2026Watched indicator—Still openSecondary indicator: if CBRT reserve data on 5 November show net reserves excluding swaps below $30 billion, room for a December rate cut narrows.
4 November 2026Falsifier—Still openFalsifier for the confluence reading: if Swiss Re or Munich Re clearly expand cyber and trade credit capacity in their nine-month results, the case that non-transferable risk is staying on balance sheets weakens.
4 November 2026Scenario%55Partly borne outH1 · 55% — Managed acceleration: Brent stays above $100, the CBRT holds monthly depreciation in a 1.8–2.0% band and reserves drain by $1–4 billion a week.
4 November 2026Scenario%30Still openH2 · 30% — Energy relief: January Brent futures fall below $90, November energy imports drop markedly and the CBRT slows the monthly pace to 1.6%.
4 November 2026Scenario%15Still openH3 · 15% — Dollarisation wave: net reserves excluding swaps fall below $30 billion, the monthly pace exceeds 2% and resident demand for foreign currency rises.
4 November 2026Causal chain—ConfirmedStep 2: as financing needs are met from reserves, gross reserves keep falling and net reserves excluding swaps approach the $30 billion threshold (lag: days).
4 November 2026Watched indicator—Still openThreshold: if Türkiye's 5-year CDS rises from 255.66 on 30 October to above 270, the market has started pricing the reserve drain into the risk premium.
3 November 2026Watched indicator—Still openThe secondary indicator is CBRT reserve data on 5 November: if net reserves excluding swaps fall below $30 billion, the 10 December rate cut narrows.
3 November 2026Watched indicator—Still openThe primary indicator is the Hormuz war premium and the pace of incidents. If January Brent closes below $100 for 5 sessions while the premium stays at 6–10%, the market is treating the constraint as overcome before we do.
3 November 2026Scenario%55Partly borne outH1 · 55% — Plan on hold, gate shut: the US does not respond to the two-day plan and attacks keep their pace. The war premium stays at 6–10%, and January Brent recovers part of its 2 November drop.
3 November 2026Scenario%30Still openH2 · 30% — A deal, but cover lags: the US and Iran agree on the Qatari plan and the strait reopens, but the Joint War Committee keeps its Gulf listing for weeks.
3 November 2026Scenario%15Still openH3 · 15% — Escalation from the Qeshm line: attacks move beyond Qeshm, and the East–West pipeline or Red Sea facilities are hit again.
3 November 2026Falsifier—Still openSuppose inventory sub-indices in the November PMIs for India and the euro area fall below 50, the US ISM prices component drops below 70 and Pakistan's wholesale inflation falls below 10%. Then the reading that costs are being deferred through inventories is refuted.
3 November 2026Scenario%55Partly borne outH1 · 55% — Sizes fixed, language loosens: the $125 billion package is kept on 4 November, and the 'at least the next several quarters' wording is replaced with something more flexible.
3 November 2026Scenario%30FalsifiedH2 · 30% — A signal of cuts at the far end: the Treasury indicates that 20- and 30-year auctions will shrink in 2027.
3 November 2026Scenario%15FalsifiedH3 · 15% — Coupon increases come early: the Treasury brings forward language on increases in the 2–10 year sector to limit the bill share.
3 November 2026Watched indicator—Still openThresholds: the 30-year above 5.40% and the 10-year above 5% would show that the coupon lock cannot protect the far end and that the term premium is widening again.
3 November 2026Causal chain—Still openIf the Fed hikes on 8–9 December, bill rates reprice within weeks; the interest bill rises without the lag that applies to debt at extended maturities.
3 November 2026Causal chain—Still openIf the 4 November statement keeps the $125 billion package, most of the first-quarter increase falls on bills and the bill share rises above 22%.
31 October 2026Scenario%50Partly borne outH1 · 50% — Scarcity stays at the front of the curve: Iran gives a partial response, the decision slips past 3 November and OPEC+ keeps its targets on 1 November.
31 October 2026Watched indicator—FalsifiedThe claim we flagged as noise: Brent's fall from 117.64 to 113.43 dollars does not show the risk premium melting; scarcity has moved to the January contract.
31 October 2026Scenario%35Still openH2 · 35% — Strike and Gulf retaliation: a US strike begins after 3 November, or limpet and Houthi attacks turn into verified damage at a facility.
31 October 2026Causal chain—Still openBreaking condition for the chain: Kpler flows return above 11 million barrels a day, the January–February spread falls below 1 dollar, or Iran responds positively to the Qatari draft.
30 October 2026Watched indicator—Still openThe primary indicator is the first independently verified damage to the state-linked fleet in the southern Hormuz corridor. If verified, the last carrier filling the space insurance has vacated comes under risk too; if not, the mine claim remains a tool for raising costs.
30 October 2026Watched indicator—Partly borne outThe secondary indicator is the January–February Brent spread after the 30 October expiry: if it stays above $3, scarcity persists at the front of the curve.
30 October 2026Scenario%55Partly borne outH1 · 55% — Balance sheets carry the load, supply continues: the state-linked fleet keeps loading without verified damage, the January–February spread stays above $3, and risk accumulates on state and corporate balance sheets.
30 October 2026Scenario%30Still openH2 · 30% — The last carrier is hit too: independently verified mine or attack damage to a state fleet tanker in the southern corridor; scarcity spreads across the whole curve.
30 October 2026Scenario%15Still openH3 · 15% — The cover door opens: Iran's 7-day plan is tied to a ceasefire and the marine war premium in Hormuz begins to fall back towards pre-war levels.
30 October 2026Falsifier—Still openIf a major exchange reopens service to Russian users by the end of November, or Yandex restores its guidance unchanged, the reading that digital payment and data infrastructure around Russia is narrowing weakens.
30 October 2026Falsifier—Still openIf war and cyber cover narrows at the January renewal while overall prices fall, the transfer of risk onto balance sheets accelerates; the cover threshold is what to watch.
30 October 2026Watched indicator—ConfirmedCalendar item: no change to December production targets is expected at the 1 November OPEC+ meeting; the real question is the capacity of the fleet carrying Gulf exports.
29 October 2026Causal chain—FalsifiedThe first step of the chain rested on the 10-year Treasury yield staying above 5%. Breaking condition: oil retreats markedly or the data weaken, pulling the 10-year below 5%.
29 October 2026Scenario%45Still openH1 · 45% — The Fed hikes in December, the BoC holds: the Fed moves to 4.00–4.25% on 9 December, while the Bank of Canada holds the same day, citing the slowdown.
29 October 2026Scenario%25Still openH3 · 25% — The Fed holds again: PCE and payrolls data weaken or oil retreats markedly, and the Fed leaves rates unchanged in December too.

Showing the latest 80 of 546 calls — all of them are in the CSV.

Published on 11 November 2026 · Reviewed on 16 November 2026

Prices signal relief, but the clock on every constraint is shortening

Open the brief ›
  • Watched indicator

    Confirmed

    The published claim

    The one thing to watch: the US 30-year auction at 21:00 on 12 November. An indirect bidder share below 70% would confirm that foreign demand is pulling back and that the term premium is widening independently of the Fed decision.

    What happened

    The threshold held by a whisker. On 12 November the Treasury sold 25 billion dollars of 30-year bonds at 5.441%. Indirect bidders took 68.1% of competitive awards, against 72.3% in October and a six-auction average of 70.5%. Dealers were left with 11.4%, and the auction tailed by 0.4 basis points. The share came in only 1.9 points below the threshold; the confirmation is weak, but the direction is right.

    Evidence: 12 November 2026 · Foreign demand fades at the US 30-year auction as dealers step in
  • Falsifier

    Partly borne out

    The published claim

    Falsifier: an extension of the US–China suspensions to 12 months at APEC in Shenzhen, together with an indirect bidder share back above 70% at the 12 November 30-year auction, would refute the 'easing on lease' reading.

    What happened

    The auction leg of the falsifier did not materialise: the indirect share did not return to 70% and stayed at 68.1%. On this leg the reading stands. The APEC leg has not yet come due; the summit is on 18–19 November. One of the two conditions has been tested and did not refute the reading; the other remains open.

    Evidence: 12 November 2026 · Foreign demand fades at the US 30-year auction as dealers step in
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Easing on lease continues: the suspensions are renewed for a short term, the Fed hikes by 25 basis points in December and Brent stays at 85–95 dollars.

    What happened

    The Fed leg strengthened. After the producer price data on 13 November, the probability of a December hike on Kalshi rose to 91.5%; CME gives 50 basis points about 10%. The Brent leg tested its limit: the futures contract closed 12 November at 95.81 dollars, just above the band. The suspension leg stays open until APEC.

    Evidence: 14 November 2026 · US producer inflation hits a three-year high; hike odds top 90%
  • Scenario

    Still open

    The published claim

    H2 · 30% — Terms lengthen and easing becomes durable: the suspensions are extended to 12 months, foreign demand returns at the 30-year auction and Hormuz traffic exceeds half of pre-war levels.

    What happened

    Two of the three conditions are moving the wrong way for now. Foreign demand did not return at the auction. Four tankers were struck in Hormuz on 10–11 November; a JMIC note of 12 November says independent AIS tracking counts transits in the single digits to low double digits per day. The suspension condition comes due on 18–19 November; the verdict stays open until then.

    Evidence: 12 November 2026 · Four tankers hit near Hormuz as Trump says no Iran deal needed

What we learned

We picked the right indicator to watch, but the margin below the threshold was 1.9 points; calling that a 'confirmation' is a stretch. From now on, wherever we set a threshold we will also state a grey band. We put the top of the Brent band at 95 dollars; the price broke through it within three days, which means we drew the band too narrow.

Published on 9 November 2026 · Reviewed on 15 November 2026

Bridges burn before Abu Dhabi as Kyiv runs short of interceptors

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 55% — Talks open, skies open: the Abu Dhabi round ends with a new schedule of talks but no ceasefire; strikes on the grid and bridges continue through the winter.

    What happened

    The horizon is the winter; six days in, two of the scenario's branches fit. The Abu Dhabi round ended on 11 November with only an agreement on a prisoner exchange of 1,000, perhaps 1,500, per side, and no ceasefire. On the first night Russia launched 162 drones and 1 ballistic missile. The third branch has not held, or is not visible: as of 14 November no date for a new round had been announced; only Kushner and Witkoff met Putin in Moscow on 12 November.

    Evidence: 12 November 2026 · Abu Dhabi talks yield a prisoner swap but leave territory unresolved
  • Scenario

    Falsified

    The published claim

    H2 · 30% — A narrow infrastructure ceasefire: the sides agree a limited ceasefire, similar to the Oleshky model, covering energy facilities and possibly ports.

    What happened

    It did not happen in the Abu Dhabi round. According to Euronews, energy infrastructure and de-escalation in the Black Sea were on the agenda of the second day on 11 November, but the round closed without producing any terms on these issues. Russia claimed to have struck a bulk carrier bound for Odesa on the night of 12 November; on 10 November it had said it hit a tanker off Odesa. The branch is closed for this round; no new claim was written for later rounds.

    Evidence: 12 November 2026 · Abu Dhabi talks yield a prisoner swap but leave territory unresolved
  • Causal chain

    Partly borne out

    The published claim

    Second link: grid damage pushes Kyiv to prioritise a narrow energy and infrastructure ceasefire in Abu Dhabi and later rounds; Moscow tries to tie this to other concessions.

    What happened

    The first half held: energy infrastructure entered the second day's agenda on 11 November. The second half could not be verified. The sources do not show which concession Moscow tied an energy ceasefire to. According to the sources, Moscow's territorial demand is that Kyiv withdraw from the eastern regions it still controls, but as of 14 November no statement showed that demand being linked to the energy track.

    Evidence: 12 November 2026 · Abu Dhabi talks yield a prisoner swap but leave territory unresolved
  • Causal chain

    Partly borne out

    The published claim

    Third link: if an infrastructure ceasefire excludes ports, port strikes keep the insurance cost of Black Sea trade high; the indicators to watch are the war-risk premium and daily Bosphorus transits.

    What happened

    No ceasefire emerged and port strikes continued. Russia said it hit 3 ships around Odesa and Chornomorsk on 8 November and a tanker on 10 November. CMA CGM has applied a war-risk surcharge of 35 dollars per TEU on all Black Sea cargo since 9 November. Data released on 12 November show the number of ships entering the Black Sea through the Bosphorus fell 41% in October. A current war-risk premium quote for voyages to Ukrainian ports could not be verified, so the link's main indicator could not be read.

    Evidence: 10 November 2026 · Russia says it hit a tanker off Odesa, two days after earlier strikes
  • Watched indicator

    Still open

    The published claim

    Threshold: January ICE Brent above 105 dollars would increase US pressure over strikes on Russian refineries and narrow Kyiv's logistical leverage.

    What happened

    The threshold was not reached. January Brent closed at 92.57 dollars on 9 November, 91.49 on 10 November, 95.81 on 12 November and 95.84 on 13 November, roughly 9 dollars short of 105. Ukraine's strikes on Russian refineries, meanwhile, accelerated; Volgograd became the eighth refinery hit since late October. Until the price threshold is crossed, the pressure channel cannot be tested.

    Evidence: Investing.com — Brent Oil Futures Historical Data
  • Falsifier

    Still open

    The published claim

    Breaking condition: if the mid-November Ukraine Defense Contact Group meeting yields new Patriot systems or interceptor deliveries, the interception rate rises and the chain breaks at the first step.

    What happened

    The horizon has not expired. According to a Ukrainska Pravda report of 29 October, the 37th meeting will be held on Tuesday 17 November at NATO headquarters in Brussels, co-chaired by the United Kingdom and Germany. The report does not disclose the agenda, and as of 14 November no record of a new Patriot delivery could be found. The verdict waits until after 17 November.

    Evidence: Ukrainska Pravda — NATO announces new Ramstein meeting on Ukraine's defence

What we learned

The report's central thesis was that the constraint lay in the sky, not at the table, and six days have supported it. Abu Dhabi delivered results only on the humanitarian track, and strikes continued even during the round. Where we were wrong was giving H2 30%. We confused energy infrastructure entering the agenda with the probability of agreement. The May 2025 Istanbul round also produced a 1,000-prisoner exchange without a ceasefire, and that precedent should have been priced in. Method note: agenda items and outcomes will be assigned separate probabilities. In the insurance link we also found no premium quote; next time the chain must be tied to an accessible indicator.

Published on 9 November 2026 · Reviewed on 14 November 2026

Oil is pricing peace while insurers still price war

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    Primary indicator: if US October CPI at 16:30 on 10 November beats the 0.3% monthly forecast, a December hike is locked in and the term premium reopens at that evening's $42 billion 10-year auction.

    What happened

    The first two links held. October CPI came in at 0.4% on the month on 10 November, with core at 0.3%. On FedWatch the probability of a 25 basis-point December hike rose from 77% to 88% in a week, and the probability of 50 basis points from zero to 11%. The Treasury sold the 10-year that evening at 5.02%; the auction tailed by 1 basis point and the bid-to-cover ratio of 2.46 was below average. The indicator's third link did not hold; details in the next review.

    Evidence: 10 November 2026 · US October inflation beats forecasts, and the same evening's auction disappoints
  • Causal chain

    Falsified

    The published claim

    The fastest channel to react to a hot US CPI will be emerging-market currencies and Türkiye's external financing costs.

    What happened

    Türkiye's 5-year CDS premium tightened after the data. According to Investing's table, it stood at 237.18 basis points on 10 November, 234.25 on 11 November and 232.93 on 12 November. Egypt's premium fell from 311.86 on 9 November to 302.45 on 11 November. The dollar index strengthened 0.67% on 11 November, but the euro took the strain, slipping to 1.1089. The CBRT's net reserves excluding swaps rose to $27.44 billion in the week to 6 November.

    Evidence: Investing.com — Turkey CDS 5 Years USD Historical Data
  • Watched indicator

    Partly borne out

    The published claim

    Secondary indicator: whether the 10–11 November Abu Dhabi round produces an energy infrastructure ceasefire clause will determine whether the Black Sea war premium drops out of its 0.6–1% band.

    What happened

    No clause emerged. The Abu Dhabi round ended on 11 November with an agreement on a prisoner exchange of 1,000, possibly 1,500, people per side. There is no provision on a ceasefire, territory or energy infrastructure. Russia claimed to have struck a tanker off Odesa on 10 November and a bulk carrier bound for Odesa on the night of 12 November. As of 13 November no new broker quote could be found showing the Black Sea war premium still in its 0.6–1% band; that half could not be verified.

    Evidence: 12 November 2026 · Abu Dhabi talks yield a prisoner swap but leave territory unresolved
  • Scenario

    Still open

    The published claim

    H1 · 50% — Prices ease, risk shifts to the public sector: Brent stays at $90–100 and Gulf political violence capacity is stuck at $200–400 million.

    What happened

    The brief set no deadline; over five trading days, events have been consistent with the scenario. Investing's table shows January Brent closing at $92.57 on 9 November, $91.49 on 10 November, $92.76 on 11 November, $95.82 on 12 November and $95.84 on 13 November. The intraday low on 10 November was $89.16; the band broke intraday but held on a closing basis. No new November-dated Marsh figure could be found for Gulf political violence capacity.

    Evidence: Investing.com — Brent Oil Futures Historical Data
  • Scenario

    Still open

    The published claim

    H2 · 30% — The gap closes in price's favour: a Washington–Tehran framework reopens Hormuz, the tanker war premium falls below 2% and Brent drops below $90.

    What happened

    The deadline has not passed, but events are running against the scenario. On 12 November Trump wrote that there was no need for a deal with Iran, and the blockade stayed in place. JMIC says 4 tankers were struck in the Strait of Hormuz and the Gulf on 10–11 November. UKMTO's 10 November report put weekly transits roughly 65% below pre-war levels. Brent did not close below $90 on any day; it settled at $95.84 on 13 November.

    Evidence: 12 November 2026 · Four tankers hit near Hormuz as Trump says no Iran deal needed
  • Scenario

    Still open

    The published claim

    H3 · 20% — The gap closes against price: an onshore Gulf facility is heavily damaged, or October CPI locks the Fed into a December hike and the dollar jumps; Brent tops $100 and CDS premiums widen.

    What happened

    One of the triggers fired, but the market outcome we expected did not follow. CPI came in at 0.4% on 10 November, a December hike became effectively fully priced, and the dollar index rose 0.67% to 102.92 on 11 November. Yet Brent closed 13 November at $95.84, and Türkiye's CDS tightened to 232.93 basis points on 12 November. There is no record of heavy damage to an onshore facility.

    Evidence: 11 November 2026 · The euro slips below 1.11 despite ECB talk of a hike
  • Falsifier

    Still open

    The published claim

    Falsifier for the combined signal: an increase in war and political violence retrocession capacity at the 1 January 2027 renewal, or Gulf political violence capacity returning above $500 million, would falsify the reading.

    What happened

    The deadline is 1 January 2027; no verdict yet. On 12 November Munich Re said it expected a tough January renewal, with prices having fallen roughly to 2021 levels; it gave no figure for war or political violence capacity. General softening does not in itself count as a falsifier. No November-dated data could be found showing Gulf capacity above $500 million.

    Evidence: 12 November 2026 · Munich Re braces for tough January renewals as rates revert to 2021

What we learned

Where we went wrong was the transmission channel. The brief said a hot US CPI would hit Türkiye's external financing costs first; the data came in hot and the auction was weak, yet Türkiye's CDS tightened from 237.18 to 232.93 basis points in three days. H3 contains the same error: we tied the macro trigger to an oil scenario, the trigger was pulled and the outcome did not follow. Method: when linking a Fed-driven shock to emerging-market risk premiums, we will spell out that week's reserve and oil direction as separate conditions.

Published on 8 November 2026 · Reviewed on 13 November 2026

The blockade moves ashore, but Iraqi oil still hangs on Hormuz insurance

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 50% — Land crossings narrow, Iraqi volumes hold: Iraq closes several crossings, sanctions hit Caspian traffic and there is no major wave of attacks in Hormuz; Iraqi exports stay at 3–3.4 million barrels a day, but conditions harden.

    What happened

    The horizon is weeks; within five days two of the branches moved the other way. On 10 November Iraq reopened the al-Sheeb crossing it had closed in October, and it is only shutting the Mandali crossing temporarily for maintenance from 13 November. Conditions eased rather than hardened: on 8 November Iraq announced an understanding with the US on Hormuz transits via a lane close to Oman. We found no record of new sanctions on Caspian traffic.

    Evidence: 8 November 2026 · Iraq agrees an Oman-side Hormuz lane for its oil with Washington
  • Scenario

    Still open

    The published claim

    H2 · 30% — A Qatari formula loosens the blockade: the US accepts a step towards lifting the blockade, Iran grants IAEA access and opens Hormuz under an interim arrangement; benchmark oil falls below 90 dollars.

    What happened

    The trigger has not been met. On 9 November Trump postponed a new strike after calls from the region, including from Qatar's prime minister, but said the blockade would continue and that Iran had not accepted the ceasefire plan. Iran proposed cutting the implementation period from 7 days to 2 while keeping its conditions. January Brent closed at 91.49 dollars on 10 November and 95.81 dollars on 12 November; no close below 90 dollars has been seen.

    Evidence: 10 November 2026 · Trump holds off on new Iran strikes but keeps the blockade
  • Scenario

    Still open

    The published claim

    H3 · 20% — Fighting resumes: the US restarts major operations before January and Iran retaliates against Gulf infrastructure; Basra loadings fall and benchmark oil rises above 110 dollars.

    What happened

    The main trigger had not been met as of 13 November; on 9 November Trump postponed new strikes. Attacks on Gulf infrastructure are continuing, however: on 9 November Saudi Arabia said the East-West pipeline had been hit in the Medina area, and the Houthis acknowledged targeting the line. Brent is well below 110 dollars, closing at 95.81 dollars on 12 November.

    Evidence: 9 November 2026 · Saudi East-West pipeline struck again near Medina
  • Watched indicator

    Still open

    The published claim

    Thresholds: January Brent above 110 dollars means a fresh Gulf outage is being priced, below 90 dollars an easing of the blockade; daily Hormuz transits below 20 mean shipowners are pulling out of Basra.

    What happened

    Neither Brent threshold was breached on a closing basis: closes from 9 to 12 November were 92.57, 91.49, 92.63 and 95.81 dollars. Sources conflict on the Hormuz count. UKMTO's weekly report of 10 November counted 170 exits and 155 entries, roughly 46 transits a day. JMIC's independent tracking, however, puts daily transits in a range from single digits to the low double digits; the bottom of that range is below the threshold.

    Evidence: 10 November 2026 · Weekly Hormuz transits remain 65% below pre-war levels
  • Causal chain

    Partly borne out

    The published claim

    Step 1 (weeks): Baghdad closes more crossings or tightens controls; Iran's western border trade is squeezed into a handful of points and tension rises with Iran-aligned Iraqi groups.

    What happened

    The political cost surfaced, but the crossings went into reverse. On 8 November Fuad Hussein said closing all of them would be difficult and that bilateral trade runs at 15–18 billion dollars a year; according to Türkiye Today, the closure had sparked protests in Maysan. Yet Iraq reopened al-Sheeb on 10 November; the only narrowing is the temporary closure of Mandali for maintenance from 13 November.

    Evidence: Türkiye Today — Iraq reopens Al-Sheeb border crossing with Iran amid US pressure
  • Causal chain

    Partly borne out

    The published claim

    Step 2 (weeks): Tehran tightens control in Hormuz, incidents in the strait multiply and the Gulf war risk premium holds in the 6–10% band or rises; buyers ask Iraq for delivery outside the strait.

    What happened

    The incident side was confirmed, the premium side was not. JMIC added 5 new incidents on 8 November, bringing the total since 1 March to 177; another tanker was hit off Oman on 10 November. Even so, Insurance Business reported on 9 November that additional war risk premiums for Gulf tankers had eased to 7–9% of hull value, with some policies written at 5%. That is despite the regional claims count exceeding 110 and estimated payouts topping 2.5 billion dollars.

    Evidence: 9 November 2026 · Gulf war premiums fall as low as 5% while losses mount
  • Causal chain

    Still open

    The published claim

    Step 3 (months): To protect revenue, Iraq leans on its own tankers and the Ceyhan line; Ceyhan loadings rise and Türkiye's transit income reflects the shift.

    What happened

    The horizon is months and a verdict would be premature. According to Windward, Kirkuk–Ceyhan loadings averaged about 341,000 barrels a day between 23 September and 28 October, below the 750,000-barrel minimum in the 1 August agreement. November data are not yet available. The Oman lane understanding and the premium falling to as low as 5% point the other way, favouring a return from Basra to the strait route.

    Evidence: 4 November 2026 · Iraq's Ceyhan exports fall short of the 750,000-barrel commitment
  • Falsifier

    Still open

    The published claim

    Breaking condition: if Qatari mediation secures an easing of the blockade and an interim reopening of Hormuz, pressure on the land crossings loses its point and the chain breaks at the first step.

    What happened

    The condition had not been met as of 13 November. Qatar's call on 9 November secured only the postponement of a new strike; on 10 November Rubio told the G7 that sanctions and the naval blockade would continue. The chain still weakened at the first step, but by another route: Washington granted Iraq the Oman lane at sea, and Baghdad reopened its land crossing.

    Evidence: 10 November 2026 · Trump holds off on new Iran strikes but keeps the blockade

What we learned

Five days on there is no scenario verdict, but we were wrong on two links of the chain. First, we assumed the premium would track the incident count; incidents and claims rose, while competition among insurers and fresh capital pushed the premium down to 5–9%. Second, we cast Iraq as a passive actor squeezed between two crossings; Baghdad reopened the crossing on 10 November. Method: in war risk premium chains we now set capacity inflows alongside the incident count.

Published on 12 September 2026 · Reviewed on 12 November 2026

The invisible front: subsea cables, zero-day vulnerabilities and 153 million identity documents

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 60% — Persistent grey-zone pressure: cable probing, espionage campaigns and data leaks continue without causing service outages; the cost builds up in defence and compliance spending.

    What happened

    In the two months since the report, events have tracked this scenario, but because it describes a continuing state and carries no deadline, a full verdict is not possible. On 22 September Volexity raised the number of groups using the same exploit kit to five. On 23 September a fire broke out at a Starlink ground station in Poland amid suspected sabotage; the station kept operating. On 22 October the UK proposed up to 14 years in prison for cable sabotage, a sign that costs are building on the compliance and legal side. As of 12 November we have no record of an incident causing a service outage in Northern Europe.

    Evidence: 22 September 2026 · Volexity: China-linked UTA0565 used a 3-flaw Chrome–Windows chain via fake sites against Asian governments; groups using the same kit rise to 5
  • Scenario

    Still open

    The published claim

    H2 · 25% — Physical outage: a data or power cable in Northern Europe is actually damaged, disrupting regional data traffic and satellite data transmission for days.

    What happened

    Not triggered as of 12 November. Between 12 September and 12 November, neither our published event records nor the search carried out for this review turned up a damaged cable in Northern Europe. The closest physical-layer incident was the 23 September fire at the Starlink ground station in Poland; it was not a cable, and it caused no outage. Because the report set no deadline, we do not count this scenario as falsified.

    Evidence: 23 September 2026 · Fire at Starlink ground station in Poland that also carries internet to Ukraine: Deputy Prime Minister Gawkowski calls it arson and sabotage
  • Causal chain

    Partly borne out

    The published claim

    Chain step 1 (days): the exploit kit circulates among groups within days and corporate patch cycles lag; commodity trading, mining and aviation companies are targeted before the vulnerability is disclosed.

    What happened

    The direction on speed was confirmed; the target forecast was not. According to Volexity's 21 September report, UTA0565 chained two Chrome vulnerabilities and one Windows vulnerability on 3–4 September; the Windows flaw was only disclosed on 8 September. The number of groups using the kit rose from four to five. But the new wave targeted government bodies in Asia, including Brunei and Japan, not commodity, mining or aviation companies.

    Evidence: 22 September 2026 · Volexity: China-linked UTA0565 used a 3-flaw Chrome–Windows chain via fake sites against Asian governments; groups using the same kit rise to 5
  • Causal chain

    Still open

    The published claim

    Chain step 3 (months): grey-zone pressure moves into the physical layer; damage to one of the two cables linking Svalbard to the mainland disrupts satellite data for days.

    What happened

    The horizon was set in months and two months have passed; there is no record of damage to the Svalbard cables. Signs that undersea pressure is continuing came from another region: on 2 November we reported Chinese research vessels slowing over Taiwanese and Japanese cables. For Northern Europe the verdict remains open.

    Evidence: 2 November 2026 · Chinese research ships linger over Taiwan's and Japan's undersea cables

What we learned

This report was written in a way that made it hard to check, and the fault is ours. None of the three scenarios carried a date or a measurable threshold, so even two months on we can only call H1 partly right and H2 and H3 open. In the first step of the chain we got the speed right but the target sector wrong: the new wave hit government bodies in Asia, not companies. The lesson is that cyber scenarios must be written with a deadline and a threshold (for example, an outage exceeding 48 hours within 90 days), and target-sector forecasts should be made only when the source supports them.

Published on 6 November 2026 · Reviewed on 11 November 2026

Prices swing while the buffers quietly drain

Open the brief ›
  • Scenario

    Still open

    The published claim

    H1 · 50% — Prices swing, buffers erode: Brent oscillates in a $90–105 band, the Gulf war premium stays at 6–10%, and the Yanbu outlet is threatened but stays open.

    What happened

    The horizon is weeks; the first two trading days fit the scenario, but the lower edge of the band was tested. According to Investing's table, January Brent (LCOF7) settled at $92.57 on 9 November and $92.29 on 10 November. The intraday low on 10 November was $89.18, so the band broke during the session but held at the close. As of 10 November no new broker data showed the war premium leaving its 6–10% band.

    Evidence: Investing.com — Brent Oil (LCOF7) historical data
  • Scenario

    Still open

    The published claim

    H2 · 30% — A political framework replenishes buffers: the war premium falls below 2%, transits rise and Egypt recalls its deferred cargoes.

    What happened

    The diplomatic precondition stirred while the buffer side moved the other way. Iraq announced an understanding with the US on the Oman lane on 8 November. The same day Egypt was reported to have cut gas to fertiliser producers for the second time; 5 of 15 contracted cargoes remain deferred to 2027, with no record of a recall. There is no data showing the premium below 2%.

    Evidence: 8 November 2026 · Egypt cuts gas to fertiliser makers again as LNG budget tightens
  • Scenario

    Still open

    The published claim

    H3 · 20% — Buffers break: the Yanbu outlet is hit or Türkiye's net reserves excluding swaps fall below $20 billion; Brent returns to its $113 peak.

    What happened

    Not triggered as of 10 November. There is no record of a strike on the Yanbu outlet; the Houthi attack of 8 November damaged the Aramco tower in Riyadh, not the port. January Brent's highest settlement between 6 and 10 November was $98.32 (Investing). The reserve threshold will be tested by the 12 November data.

    Evidence: 8 November 2026 · Houthi missiles leave damage at an Aramco tower in Riyadh
  • Watched indicator

    Still open

    The published claim

    Primary indicator: if CBRT weekly data on 12 November show net reserves excluding swaps falling from $24.3 billion to below $20 billion, the currency anchor's capacity to carry the energy bill reaches its limit.

    What happened

    Not yet due. The data will be released on 12 November at 14:30; this check was written on 11 November. The verdict will be given that day, and the $24.3 billion figure for the week of 30 October remains the last known value.

  • Falsifier

    Still open

    The published claim

    Confluence falsifier: a rise in net reserves excluding swaps on 12 November, Egypt recalling its deferred cargoes and the Gulf war premium falling below 2% would disprove the reading.

    What happened

    As of 10 November, two of the three strands are not pointing towards falsification. Egypt cut gas to industry for the second time on 8 November and did not recall cargoes. There is no data showing the war premium below 2%. The reserve strand opens on 12 November, so the verdict remains open.

    Evidence: 8 November 2026 · Egypt cuts gas to fertiliser makers again as LNG budget tightens

What we learned

Five days on there is no verdict, but we are recording a numerical error here. The brief and our sourced data gave the 6 November January Brent settlement as $97.75; Investing's table today shows $98.32, the same figure CNBC reported. The $92.31 entered for 9 November was likewise an intraday reading rather than the final settlement, which was $92.57. Both have been corrected in the sourced data. Method: we no longer enter the brent-vadeli value from a table read before the close; we re-verify the same day against the same source the following night.

Published on 5 November 2026 · Reviewed on 10 November 2026

The election sank oil, but only three ships crossed the strait

Open the brief ›
  • Scenario

    Still open

    The published claim

    H1 · 55% — Prices fall, the strait stays shut: Iran does not narrow its blacklist, daily transits stay at 3–7 ships, the war-risk premium holds in the 6–10% band, and part of the drop to 94.62 dollars is recovered within weeks.

    What happened

    The horizon is weeks and there is no verdict yet; the first five days give mixed signals. January Brent rose to 98.29 dollars on 5 November, recovering part of the drop, but fell to 92.31 dollars on 9 November (Investing, LCOF7). The strait stayed shut: a tanker caught fire off Hasab on 7 November, and Iran's claim that it struck a mine could not be verified. As of 9 November, there has been no announcement that the blacklist has narrowed.

    Evidence: 7 November 2026 · Tanker burns off Khasab as Iran's mine claim goes unverified
  • Scenario

    Still open

    The published claim

    H2 · 30% — Political framework, gradual reopening: the new balance in Washington speeds up framework talks with Iran, the blacklist narrows and transits with tracking signals on return to double digits.

    What happened

    Only the scenario's diplomatic precondition has moved. On 8 November Iraq announced an understanding with the US on an Omani lane for its Hormuz exports; the lane's coordinates and Iran's role were not disclosed. On the same day Araghchi said the strait could reopen within 2 days if conditions were met. As of 9 November, it could not be verified that the blacklist had narrowed or that daily transits had returned to double digits.

    Evidence: 8 November 2026 · Iraq agrees an Oman-side Hormuz lane for its oil with Washington
  • Scenario

    Still open

    The published claim

    H3 · 15% — The list turns into seizures: Iran seizes one of the listed Gulf state tankers or stops vessels conducting ship-to-ship transfers; January Brent returns above 100 dollars.

    What happened

    As of 9 November, there is no record of a listed state tanker being seized. January Brent has not returned above 100 dollars; its highest close between 5 and 9 November was 98.32 dollars (Investing). The event closest to the trigger was the Hasab fire on 7 November, and its cause could not be verified.

  • Watched indicator

    Still open

    The published claim

    Primary indicator: if January Brent falls below 90 dollars while transits stay at 3–7 ships, the market is treating the constraint as overcome before we do.

    What happened

    The threshold was approached but not breached. According to Investing data, January Brent fell intraday to 90.67 dollars on 9 November and closed at 92.31 dollars; the same table gives 98.32 dollars for 6 November. Trading Economics gives 93.38 dollars for the same day, as the two sources differ on closing time. No independent data could be found showing a change in daily transit numbers over this period.

    Evidence: Investing.com — Brent Oil (LCOF7) historical data
  • Watched indicator

    Still open

    The published claim

    Secondary indicator: the US affiliates rule and China's rare earth suspensions on 10 November; if no notice is published, access shifts from a contractual term to a formal restriction.

    What happened

    Half of it is resolved. On 5 November the US published a text extending the suspension of the affiliates rule to 10 January 2027, so the restriction did not become formal on that side. As of the morning of 10 November, no new Chinese text carrying the suspension in Announcement No. 70 beyond 10 November could be found. The verdict depends on Beijing's announcement.

    Evidence: 5 November 2026 · Washington extends its affiliates rule pause as Beijing stays silent
  • Falsifier

    Still open

    The published claim

    Falsifier for the converging-signals reading: the reading weakens if the US and Chinese suspensions are extended by notice on 10 November, if the ex-China price of yttrium halves by the end of December, or if daily transits with tracking signals on return to double digits in Hormuz.

    What happened

    As of the morning of 10 November, none of the three branches has been triggered. The US wrote its extension, China did not; the end-December deadline for the yttrium price has not yet arrived. Double-digit transits with tracking signals on in Hormuz could not be verified as of 9 November.

What we learned

Five days on, no scenario can be ruled on; their horizons are weeks. The warning sign is in the price. The brief said the drop would be partly recovered within weeks; January Brent rallied for two days, then fell to 92.31 dollars on 9 November with no verified reopening of the strait. As on 4 November, the price may again be pricing diplomatic headlines ahead of physical transits; the cause of the 9 November drop could not be verified with a source. Method: we keep the 90-dollar threshold, but will not issue a verdict without matching Brent and transit counts from the same day's data. We also record the closing-price gap between the two sources separately.

Published on 5 November 2026 · Reviewed on 10 November 2026

The Hormuz counter shows three ships as Gulf oil flows dark

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 50% — The dark corridor persists: the blockade and the PGSA list stay in place, transits with tracking signals on stay below 10 a day, and Gulf exports hold in the 15–18 million barrels a day band.

    What happened

    As of 9 November the corridor has not opened, but the volume side strengthened in Iraq. Iraq's oil minister put October exports at 3.4 million barrels a day, the pre-war level; about 93% of that still transits the strait from Basra. Vortexa data showing the November Gulf total are not yet available.

    Evidence: 7 November 2026 · Iraqi exports are back to pre-war levels, but Hormuz still carries them
  • Scenario

    Still open

    The published claim

    H2 · 30% — Deal and gradual reopening: the US and Iran reach an interim deal easing the blockade and sanctions, the PGSA list is suspended, and premiums fall with a lag as insurers wait for an incident-free period.

    What happened

    As of 9 November there is no interim deal. There is a step in the opposite direction: on 7 November the US set up a task force extending the Iran blockade to land borders and the Caspian. Partial easing appears only for Iraq; an understanding with the US on an Omani lane was announced on 8 November, and the PGSA list has not been suspended.

    Evidence: 7 November 2026 · US task force extends Iran blockade to land borders and the Caspian
  • Scenario

    Still open

    The published claim

    H3 · 20% — The list cuts the transfer chain: the PGSA lists foreign tankers conducting transfers and seizes one, or an attack occurs; Gulf exports fall below 15 million barrels a day.

    What happened

    As of 9 November there is no record of a transfer tanker being seized. The attack branch is unclear: a tanker caught fire off Hasab on 7 November, and the Revolutionary Guards claimed that a tanker with its tracking signal off had struck a mine, which could not be independently verified.

    Evidence: 7 November 2026 · Tanker burns off Khasab as Iran's mine claim goes unverified
  • Watched indicator

    Still open

    The published claim

    Threshold: if January Brent falls below 90 dollars while the blacklist grows, futures prices have decoupled from the delivery constraint.

    What happened

    The threshold was tested on 9 November but not breached. According to Investing data, January Brent fell intraday to 90.67 dollars and closed at 92.31 dollars. On the day the report was published it stood at 98.29 dollars; it fell 6.1% over four sessions.

    Evidence: Investing.com — Brent Oil (LCOF7) historical data
  • Causal chain

    Partly borne out

    The published claim

    Step one (days): listed state fleets turn to transits with tracking signals off and to ship-to-ship transfers at sea.

    What happened

    State fleets were seen taking the risk onto their own hulls, but the count of dark transits could not be updated. Iraq's state tanker company moved 2 million barrels through the strait on its own supertanker, a first for SOMO in decades. Whether the transit was made with the tracking signal off, and Windward's transfer count after 5 November, could not be verified.

    Evidence: 7 November 2026 · Iraqi exports are back to pre-war levels, but Hormuz still carries them
  • Causal chain

    Still open

    The published claim

    Step two (weeks): the PGSA also lists foreign vessels taking over transfers, the war-risk premium does not fall below the 6–10% band, and the transfer link becomes more expensive.

    What happened

    The horizon is weeks. As of 9 November, no PGSA update listing foreign transfer tankers could be found. The latest broker data we have for the war-risk premium is still the 6–10% band from 14 October; no more recent rate could be verified with a source.

What we learned

Five days on, the report is mostly open; the dark corridor persists and no branch has resolved. What we missed was Iraq. The report said state fleets would carry the risk through transits with tracking signals off and through transfers; Iraq instead transited openly with a state tanker and secured a route understanding from the US. In other words, the constraint is easing for one country through political bargaining while the list stays in place. Method: we will track Gulf exporters not as a single bloc but individually, as countries able to bargain separately with Washington.

Published on 4 November 2026 · Reviewed on 9 November 2026

Cheaper oil leaves the bill for the shock on balance sheets

Open the brief ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — The window closes and the blockade holds: Qatar's 36-hour window lapses unanswered, tanker attacks continue, the war premium stays at 6–10% and January Brent recovers part of its two-session fall.

    What happened

    The constraint leg held; the price leg did not. As of 9 November no framework had been announced, and on 7 November the US set up the Iron Gate task force, extending the blockade to land borders and the Caspian. January Brent, rather than recovering, fell 8.7% to $94.62 on 4 November. It closed at $97.75 on 6 November, still below the $103.60 of 3 November.

    Evidence: 7 November 2026 · US task force extends Iran blockade to land borders and the Caspian
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Framework deal, delayed cover: the US and Iran reach a framework on the Qatari proposal and the strait reopens, but the Joint War Committee keeps its Gulf listing for weeks.

    What happened

    The 36-hour window passed without a framework and the strait did not reopen. The Revolutionary Guards claimed on 5 and 6 November that tankers using unauthorised routes had struck mines; maritime officials did not confirm this. Prices, though, behaved as this scenario predicted: January Brent closed below $100 for 3 sessions.

    Evidence: 6 November 2026 · Revolutionary Guards claim two more tankers hit mines, with no confirmation
  • Scenario

    Falsified

    The published claim

    H3 · 15% — The window closes in conflict: attacks extend beyond Qeshm, or the Habshan–Fujairah line and Red Sea facilities are hit, and cover is withdrawn under 7-day cancellation clauses.

    What happened

    The window did not close in conflict, and no strikes were recorded on the lines that bypass Hormuz. The closest development to the trigger came on 5 November, when Saudi civil defence issued a danger alert for Yanbu and 1 ballistic missile fired at Medina was intercepted.

    Evidence: 5 November 2026 · Saudi Arabia issues a danger warning for Yanbu, its Hormuz bypass outlet
  • Watched indicator

    Falsified

    The published claim

    Primary indicator: if January Brent falls below $100 while the window lapses unanswered, the market has moved past the constraint before us; Brent at $103.60 does not translate into physical supply until the war premium drops below 2%.

    What happened

    Our own test proved us wrong. As the window passed without a framework, Brent fell to $94.62 on 4 November. CNBC attributed the drop to the midterm result and to Trump saying he had no urgent plan to strike Iran. The price settled below $100 with no easing at the insurance gate, closing at $98.29 on 5 November and $97.75 on 6 November.

    Evidence: CNBC — Oil close on 4 November 2026 and the midterm election effect
  • Watched indicator

    Still open

    The published claim

    Secondary indicator: if CBRT reserve data on 5 November show net reserves excluding swaps below $30 billion, room for a December rate cut narrows.

    What happened

    The threshold was crossed; the outcome is not yet visible. CBRT data show net reserves excluding swaps falling by $6.5 billion to $24.3 billion in the week of 30 October, while foreigners sold $749 million of Turkish government bonds the same week. The 10 December decision will show whether the room for a cut has narrowed.

    Evidence: 5 November 2026 · Net reserves excluding swaps hit a seven-month low as foreigners exit bonds
  • Falsifier

    Still open

    The published claim

    Falsifier for the confluence reading: if Swiss Re or Munich Re clearly expand cyber and trade credit capacity in their nine-month results, the case that non-transferable risk is staying on balance sheets weakens.

    What happened

    The Swiss Re half did not trigger the falsifier. Rather than expanding capacity, the company said on 5 November that it had cut property and casualty reinsurance revenue from $14 billion to $12.8 billion, with the new-business contract margin down about 28%. Munich Re reports third-quarter results on 12 November; the verdict stays open until then.

    Evidence: 5 November 2026 · Swiss Re lets P&C reinsurance revenue shrink by 1.2 billion dollars

What we learned

Once again we went wrong where price and insurance cover diverged, and this time we had written it into our own threshold. The brief defined Brent falling below $100 while the window lapsed without a framework as the moment the market overtook us; that is exactly what happened on 4 November. The trigger was not Hormuz but the US midterm election, so the price was reflecting Washington's cost of continuing the war, not physical supply. None of our scenarios included that political channel. Method: we will add a US domestic-politics trigger to Gulf scenarios as a separate branch, and assign separate probabilities to price and to insurance cover.

Published on 4 November 2026 · Reviewed on 9 November 2026

The lira passes 50 as the energy bill loosens the crawl

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Managed acceleration: Brent stays above $100, the CBRT holds monthly depreciation in a 1.8–2.0% band and reserves drain by $1–4 billion a week.

    What happened

    The scenario's trigger broke down within 2 days. January Brent closed at $94.62 on 4 November and $97.75 on 6 November, staying below $100. The reserve leg, however, moved faster than expected: gross reserves fell by $4.1 billion in the week of 30 October and net reserves excluding swaps by $6.5 billion. The end-November level of the lira and the monthly pace cannot yet be measured.

    Evidence: 5 November 2026 · Net reserves excluding swaps hit a seven-month low as foreigners exit bonds
  • Scenario

    Still open

    The published claim

    H2 · 30% — Energy relief: January Brent futures fall below $90, November energy imports drop markedly and the CBRT slows the monthly pace to 1.6%.

    What happened

    Brent turned in this scenario's direction but did not reach the threshold. The lowest close was $94.62 on 4 November; no print below $90 has been seen. Trade data showing November energy imports will be published in early December.

    Evidence: CNBC — Oil close on 4 November 2026 and the midterm election effect
  • Scenario

    Still open

    The published claim

    H3 · 15% — Dollarisation wave: net reserves excluding swaps fall below $30 billion, the monthly pace exceeds 2% and resident demand for foreign currency rises.

    What happened

    Two of the three triggers showed up in the 5 November data. Net reserves excluding swaps fell to $24.3 billion, and residents' FX deposits, adjusted for parity effects, rose by $585 million in the same week. Whether the monthly pace has exceeded 2%, and whether the authorities respond through withholding tax or rates, is not yet visible.

    Evidence: 5 November 2026 · Net reserves excluding swaps hit a seven-month low as foreigners exit bonds
  • Causal chain

    Confirmed

    The published claim

    Step 2: as financing needs are met from reserves, gross reserves keep falling and net reserves excluding swaps approach the $30 billion threshold (lag: days).

    What happened

    The step played out on the expected timetable and more sharply than forecast. According to the 5 November data, net reserves excluding swaps dropped to $24.3 billion in the week of 30 October, from $39.34 billion in the week of 16 October. Gross reserves slipped to $158.1 billion.

    Evidence: 5 November 2026 · Net reserves excluding swaps hit a seven-month low as foreigners exit bonds
  • Watched indicator

    Still open

    The published claim

    Threshold: if Türkiye's 5-year CDS rises from 255.66 on 30 October to above 270, the market has started pricing the reserve drain into the risk premium.

    What happened

    CDS moved the other way. On 5 November, the day the reserve data came out, the 5-year CDS fell to 246.79, from 257.22 on 3 November and 251.35 on 4 November. For now the market is not pricing the reserve drain into the risk premium.

    Evidence: Investing.com — Turkey CDS 5 Years USD Historical Data

What we learned

The report's mechanism held on the reserve side but not on the trigger side. We tied the reserve drain to the energy bill and conditioned it on Brent staying above $100; Brent fell below $100 two days later, yet reserves still drained by $6.5 billion. Foreign sales of $749 million in Turkish government bonds the same week suggest that part of the drain came through the portfolio channel rather than energy. The 15% we gave H3 proved too low, since two of its triggers materialised within two days. The fall in CDS does not, for now, support our assumption that reserve losses would feed into the risk premium. Method: in reserve theses we will now write the energy and portfolio channels as separate triggers.

Published on 3 November 2026 · Reviewed on 8 November 2026

Oil eased in a day; war cover stays shut for weeks

Open the brief ›
  • Watched indicator

    Still open

    The published claim

    The secondary indicator is CBRT reserve data on 5 November: if net reserves excluding swaps fall below $30 billion, the 10 December rate cut narrows.

    What happened

    The threshold was crossed; we have not yet seen the outcome. Data cited by Bloomberg HT at 14:46 on 5 November show net reserves excluding swaps falling from $30.8 billion to $24.3 billion in the week of 30 October. In the same week net reserves fell from $42.3 billion to $38 billion. Whether the cut narrows will be seen in the 10 December decision.

    Evidence: 5 November 2026 · Net reserves excluding swaps hit a seven-month low as foreigners exit bonds
  • Watched indicator

    Still open

    The published claim

    The primary indicator is the Hormuz war premium and the pace of incidents. If January Brent closes below $100 for 5 sessions while the premium stays at 6–10%, the market is treating the constraint as overcome before we do.

    What happened

    January Brent closed at $94.62 on 4 November, $98.29 on 5 November and $97.75 on 6 November. So 3 of the 5 sessions in the threshold have been met, with the remaining 2 on 9 and 10 November. There is no easing on the premium side: 35 ships were hit in October, and at least 5 incidents were reported to UKMTO between 1 and 5 November. Brokers' last public reading dates from 14 October and is 6–10% of hull value.

    Evidence: 5 November 2026 · October was the war's worst month for ship strikes, and fires continue
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Plan on hold, gate shut: the US does not respond to the two-day plan and attacks keep their pace. The war premium stays at 6–10%, and January Brent recovers part of its 2 November drop.

    What happened

    The constraint side held; the price side did not. As of 7 November the plan had not been accepted, and the same day the US set up a task force extending the blockade to land borders and the Caspian. After 35 hits in October, at least 5 incidents were recorded in the first 5 days of November. But January Brent did not recover from $104.77 on 2 November; it fell to $94.62 on 4 November and stood at $97.75 on 6 November.

    Evidence: 7 November 2026 · US task force extends Iran blockade to land borders and the Caspian
  • Scenario

    Still open

    The published claim

    H2 · 30% — A deal, but cover lags: the US and Iran agree on the Qatari plan and the strait reopens, but the Joint War Committee keeps its Gulf listing for weeks.

    What happened

    We set no deadline for this scenario. As of 7 November there is no agreement. According to a CNBC report on 5 November, Eurasia Group cut the probability of a deal from 20% to 10% and raised the probability of a continued stalemate from 15% to 30%.

    Evidence: 5 November 2026 · October was the war's worst month for ship strikes, and fires continue
  • Scenario

    Still open

    The published claim

    H3 · 15% — Escalation from the Qeshm line: attacks move beyond Qeshm, and the East–West pipeline or Red Sea facilities are hit again.

    What happened

    The trigger came close to the threshold but did not occur. On 5 November Saudi civil defence issued a danger warning for Yanbu, the terminus of the line that bypasses Hormuz; 1 ballistic missile fired at Medina was destroyed in the air. Houthi claims about Riyadh and Dammam airports were not verified. As of 7 November there is no verified damage to the East–West line or Red Sea facilities.

    Evidence: 5 November 2026 · Saudi Arabia issues a danger warning for Yanbu, its Hormuz bypass outlet
  • Falsifier

    Still open

    The published claim

    Suppose inventory sub-indices in the November PMIs for India and the euro area fall below 50, the US ISM prices component drops below 70 and Pakistan's wholesale inflation falls below 10%. Then the reading that costs are being deferred through inventories is refuted.

    What happened

    The deadline is the November data, in the first week of December. As of 8 November there are no new data to rule on. The latest readings are for October: the US ISM prices component at 75.2, India's manufacturing PMI at 54.7 and Pakistan's wholesale inflation at 13.8%.

What we learned

In this brief we fell short in the same place again: we saw the constraint correctly but missed the price. The war premium and attack-pace parts of H1 held, but none of the 3 scenarios foresaw that January Brent could fall below $100 while the cover gate stayed shut. Brent closed below $100 for 3 sessions on 4–6 November. Our secondary indicator, meanwhile, proved measurable and crossed its threshold by a margin of $5.7 billion. Method: we will write the divergence between price and cover as a separate scenario, and put dates on triggers.

Published on 3 November 2026 · Reviewed on 8 November 2026

Treasury's coupon lock pushes rate risk onto bills

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Sizes fixed, language loosens: the $125 billion package is kept on 4 November, and the 'at least the next several quarters' wording is replaced with something more flexible.

    What happened

    The package held; the language call did not. In its 4 November statement the Treasury said it would refinance about $97 billion maturing on 15 November with $125 billion of auctions. But it repeated verbatim the wording that coupon sizes would be maintained for at least the next several quarters. The only structural change was that the 7-year note will move from monthly to quarterly new issuance from March 2027.

    Evidence: 4 November 2026 · US Treasury moves seven-year notes from monthly to quarterly new issues
  • Scenario

    Falsified

    The published claim

    H2 · 30% — A signal of cuts at the far end: the Treasury indicates that 20- and 30-year auctions will shrink in 2027.

    What happened

    The 4 November statement gives no sign that 20- and 30-year auctions will shrink; the $25 billion 30-year auction on 12 November stands as planned. On the contrary, the Treasury Borrowing Advisory Committee wrote that coupon increases would be needed late in fiscal 2027, with larger ones in 2028 and 2029.

    Evidence: 4 November 2026 · US Treasury moves seven-year notes from monthly to quarterly new issues
  • Scenario

    Falsified

    The published claim

    H3 · 15% — Coupon increases come early: the Treasury brings forward language on increases in the 2–10 year sector to limit the bill share.

    What happened

    The Treasury's own statement did not bring forward any language on increases; it repeated that sizes would be maintained for at least the next several quarters. Only the advisory committee report mentioned a timetable for increases, and it placed them late in fiscal 2027. The committee noted that current sizes would shift the debt towards bill-heavy, nearer maturities; this supports the report's main thesis.

    Evidence: U.S. Department of the Treasury — Quarterly Refunding Statement, November 4, 2026
  • Watched indicator

    Still open

    The published claim

    Thresholds: the 30-year above 5.40% and the 10-year above 5% would show that the coupon lock cannot protect the far end and that the term premium is widening again.

    What happened

    We set no deadline for the thresholds; as of 5 November neither had been crossed, and the far end eased. FRED data show the 30-year constant maturity yield falling from 5.32% on 3 November to 5.25% on 5 November, and the 10-year from 4.95% to 4.84%. In its first 2 days the statement added no supply worries at the far end, but the 10 and 12 November auctions have yet to take place.

    Evidence: FRED — Market Yield on U.S. Treasury Securities at 30-Year Constant Maturity (DGS30)
  • Causal chain

    Still open

    The published claim

    If the Fed hikes on 8–9 December, bill rates reprice within weeks; the interest bill rises without the lag that applies to debt at extended maturities.

    What happened

    The deadline is the 9 December decision. The precondition has strengthened: October payrolls came in at 165,000 on 6 November, and the priced probability of a December hike rose to 77%. When the report was written it was 73%.

    Evidence: 6 November 2026 · US payrolls rise by nearly three times the forecast in October
  • Causal chain

    Still open

    The published claim

    If the 4 November statement keeps the $125 billion package, most of the first-quarter increase falls on bills and the bill share rises above 22%.

    What happened

    The condition was met: the package was kept at $125 billion. The monthly debt statement that would show whether the bill share has passed 22% has not yet been published. The advisory committee wrote that current sizes would shift the debt towards bill-heavy, nearer maturities; that direction fits the thesis, but the ratio has not been verified.

    Evidence: 4 November 2026 · US Treasury moves seven-year notes from monthly to quarterly new issues

What we learned

The report's main thesis held: the Treasury locked in coupons, and the shift to nearer maturities made it into the advisory committee's own text. Where we were wrong was on language. In the most likely scenario we wrote that the 'at least the next several quarters' wording would loosen; the Treasury repeated it verbatim. The only unexpected move, the switch of the 7-year to quarterly issuance, appeared in none of our scenarios. Method: in refunding statements we will track changes to the auction calendar and series structure, not changes in wording.

Published on 22 October 2026 · Reviewed on 7 November 2026

Britain's rate rise is settled, but the budget has no energy headroom

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — One hike, a cautious budget: the BoE raises Bank Rate to 4% on 5 November and the 28 October budget keeps energy support limited. Gilt yields hold around 5.4%, inflation peaks around 4% in early 2027 and the committee waits until February.

    What happened

    Both triggers materialised. On 5 November the BoE raised Bank Rate from 3.75% to 4% by 8 votes to 1, and Danske judged the budget a cautious one that added little to near-term borrowing. Two of the outcomes did not. The ten-year gilt closed at 5.15% on 5 November, 25 basis points below 5.4%. The BoE expects inflation to reach about 4.5% in the first quarter of 2027 and about 5% in the second; we had written of a peak around 4%. The wait-until-February element is still open.

    Evidence: 5 November 2026 · Bank of England raises rates to 4% on the energy shock
  • Scenario

    Still open

    The published claim

    H2 · 30% — Fiscal expansion, a second hike: the budget funds energy support with extra borrowing, October inflation exceeds 3.75% and the BoE moves to a second hike before February.

    What happened

    The first trigger did not materialise: according to the 3 November assessment the budget added little to near-term borrowing, and the BoE minutes say budget measures will lower CPI by 0.25 points until the second quarter of 2027. October inflation has not yet been published. Bailey said the 5 November hike would not be the last if the conflict is not resolved; the February meeting is the one to wait for.

    Evidence: 3 November 2026 · Markets fully price a BoE hike, shifting debate to February
  • Scenario

    Falsified

    The published claim

    H3 · 15% — Growth cracks, the hike is postponed: budget tightening and weak data push the committee to hold in November, leaving Bank Rate at 3.75%.

    What happened

    The committee did not hold on 5 November. Bank Rate rose to 4%, and of the nine members only Catherine Mann dissented, voting for a 50 basis point move to 4.25%. No member argued for a hold.

    Evidence: 5 November 2026 · Bank of England raises rates to 4% on the energy shock
  • Causal chain

    Partly borne out

    The published claim

    Link 1: the 28 October budget is forced to keep energy support limited; if support is funded with extra borrowing, gilt issuance rises and the ten-year yield is pushed above roughly 5.4%.

    What happened

    The constraint held; the market reaction did not. The budget stayed cautious and added little to borrowing. Yields, however, were not pushed to 5.4%: the ten-year gilt closed at 5.24% on 3 November and 5.15% on 5 November, down about 15 basis points in a month. The pressure stayed at the 30-year tenor, which rose to 5.63% on 5 November.

    Evidence: 3 November 2026 · Markets fully price a BoE hike, shifting debate to February
  • Causal chain

    Partly borne out

    The published claim

    Link 2: a higher Bank Rate and higher yields at the far end of the curve lift mortgage rates, the two-year fixed rate climbs further and mortgage approvals fall.

    What happened

    The BoE says the two-year fixed mortgage rate has risen by about 50 basis points since early September and household mortgage approvals are falling. The direction is right, but most of the rise came before our report of 22 October; there are no November data yet to isolate the effect of the hike itself.

    Evidence: Bank of England — Monetary Policy Summary and minutes, November 2026
  • Watched indicator

    Still open

    The published claim

    Threshold: Brent futures settling durably above 110 dollars would show fuel-driven inflation combining with the January bill increase to push the BoE into a second hike.

    What happened

    Brent futures settled above 110 dollars in seven sessions between 22 and 30 October, peaking at 116.72 dollars on 23 October. They then fell below the threshold: 104.77 dollars on 2 November, 94.62 dollars on 4 November and 97.75 dollars on 6 November. The move did not prove durable; because no deadline was set for the threshold, the verdict stays open.

    Evidence: Investing.com — Brent Oil Futures Historical Data
  • Watched indicator

    Still open

    The published claim

    Threshold: EUR/USD falling below its mid-October low of 1.11 would show a France-driven periphery premium passing into the currency.

    What happened

    Not breached. In FRED's daily series EUR/USD stood at 1.1212 on 22 October, hit a low of 1.1140 on 28 October and closed at 1.1163 on 30 October. It came within 40 pips of the threshold at most; because no deadline was set, the verdict stays open.

    Evidence: FRED — U.S. Dollars to Euro Spot Exchange Rate (DEXUSEU)
  • Watched indicator

    Still open

    The published claim

    Threshold: the US 30-year yield exceeding 5.6% would show a global term premium intensifying pressure on gilts and OATs and narrowing fiscal room.

    What happened

    Not breached. The 30-year yield was 5.29% on 22 October, the day the report was published; its high for the period was 5.37% on 28 October, and it fell to 5.25% on 5 November. That is 35 basis points below the threshold; because no deadline was set, the verdict stays open.

    Evidence: FRED — 30-Year Treasury Constant Maturity Rate (DGS30)

What we learned

We read the direction correctly and the magnitude wrongly. The hike and the cautious budget came, and H3 was falsified. But we got two numbers wrong: gilts fell to 5.15% rather than holding at 5.4%, and the inflation peak is now forecast at about 5%, not 4%. We assumed the fiscal constraint would push yields up; with the budget staying cautious, pressure left the ten-year and moved to the 30-year. We also underestimated the pass-through of energy into inflation. Lesson: state currency and yield outcomes as claims separate from their triggers, and calculate energy pass-through using the institution's own assumed price. H2 and the thresholds will be audited again after October CPI and the February decision.

Published on 29 October 2026 · Reviewed on 6 November 2026

Canada's hike flattened its curve; the Fed's hold steepened America's

Open the report ›
  • Causal chain

    Falsified

    The published claim

    The first step of the chain rested on the 10-year Treasury yield staying above 5%. Breaking condition: oil retreats markedly or the data weaken, pulling the 10-year below 5%.

    What happened

    The breaking condition was met within a week. January Brent fell from 113.43 dollars on 30 October to 103.60 dollars on 3 November. On the Treasury's constant-maturity curve the 10-year eased from 5.036% on 28 October to 4.95% on 3 November. On 4 November the yield dropped to 4.90% and front-month Brent to 94.62 dollars; the chain snapped at the first link.

    Evidence: 3 November 2026 · Treasury yields climb on election day ahead of the refunding statement
  • Scenario

    Still open

    The published claim

    H1 · 45% — The Fed hikes in December, the BoC holds: the Fed moves to 4.00–4.25% on 9 December, while the Bank of Canada holds the same day, citing the slowdown.

    What happened

    The scenario falls due on 9 December, 33 days away. In the meantime the energy premium it relied on has thinned: front-month Brent fell 8.7% in a single day on 4 November to 94.62 dollars. We found no sourced data on how the probability of a December hike has moved since the 82% of 28 October.

    Evidence: CNBC — Oil prices, 4 November 2026
  • Scenario

    Still open

    The published claim

    H3 · 25% — The Fed holds again: PCE and payrolls data weaken or oil retreats markedly, and the Fed leaves rates unchanged in December too.

    What happened

    The oil leg of the trigger was met by 4 November: front-month Brent fell 16.6% between 30 October and 4 November. The decision is on 9 December, and the October payrolls report due on 6 November had not been released when this check was written. We leave the probability at 25% as published.

    Evidence: 3 November 2026 · Treasury yields climb on election day ahead of the refunding statement

What we learned

We read the direction of the curve correctly but got the chain's foundation wrong. On 28 October the market priced the Fed's pause as an inflation risk and the 10-year rose to 5.036%; that reading still holds a week later. But we tied the housing and emerging-market issuance chain to the 10-year staying above 5%, and the yield fell to 4.90% on 4 November. We wrote a marked retreat in oil only as the trigger for H3, at 25%; front-month Brent fell 16.6% in a week. Method: we will anchor chain foundations to a range rather than a single threshold, and for energy-sensitive thresholds we will state oil's weekly volatility separately.

Published on 31 October 2026 · Reviewed on 6 November 2026

The roll did not ease anything; scarcity moved into January Brent

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Scarcity stays at the front of the curve: Iran gives a partial response, the decision slips past 3 November and OPEC+ keeps its targets on 1 November.

    What happened

    All three triggers were met. On 1 November the seven core OPEC+ members froze their December target at about 31 million barrels a day. On 3 November the Iranian official Rezaei said the strait would not reopen until seven conditions were met; Qatar was waiting 36 hours for a reply. Price, however, moved against the scenario: January Brent fell from 113.43 dollars on 30 October to 103.60 dollars on 3 November.

    Evidence: 1 November 2026 · OPEC+ freezes December quotas with effective spare capacity at 270,000 barrels
  • Watched indicator

    Falsified

    The published claim

    The claim we flagged as noise: Brent's fall from 117.64 to 113.43 dollars does not show the risk premium melting; scarcity has moved to the January contract.

    What happened

    The fall we called noise became the main move over the next three sessions. According to CNBC, front-month Brent dropped 8.7% to 94.62 dollars on 4 November, and oil futures lost more than 14% over the week. The trigger was not returning supply but expectations of a diplomatic settlement after the midterms. The same report said Trump told the Financial Times there was no plan for an imminent strike on Iran; Brent's one-day fall on 4 November reached 8.7%.

    Evidence: CNBC — Oil prices, 4 November 2026
  • Scenario

    Still open

    The published claim

    H2 · 35% — Strike and Gulf retaliation: a US strike begins after 3 November, or limpet and Houthi attacks turn into verified damage at a facility.

    What happened

    We set no deadline for this scenario. On 2 November the US hit one launcher on Qeshm Island, but three US officials said it did not mark a return to major operations. On 3 November a tanker was struck by two projectiles off Lima; as of 5 November there is no verified damage at any facility.

    Evidence: 2 November 2026 · US hits launcher on Qeshm Island but says major campaign not resumed
  • Causal chain

    Still open

    The published claim

    Breaking condition for the chain: Kpler flows return above 11 million barrels a day, the January–February spread falls below 1 dollar, or Iran responds positively to the Qatari draft.

    What happened

    None of the three conditions had been confirmed by a source as of 5 November. According to MUFG, the spread between the first two contracts was 3 dollars on 2 November, 2 dollars above the 1 dollar threshold. We found no sourced spread data for 3 and 4 November. According to Windward, only three ships transited the Strait of Hormuz on 3 November.

    Evidence: 3 November 2026 · Only three ships transit Hormuz as Iran doubles its blacklist

What we learned

We read the physical constraint correctly and misread what the price was pricing. The quota, Iran's seven conditions and the Hormuz transit count still signalled scarcity on 5 November; yet front-month Brent fell from 113.43 to 94.62 dollars between 30 October and 4 November, a drop of 16.6%. Labelling that fall as noise was a mistake: the market was pricing not physical supply but the chance that US domestic politics would shorten the war. Method: we will write the political calendar (elections, budgets, war-funding votes) into scenario triggers as a separate channel, and use the noise label only once the mechanism carrying the price has been refuted.

Published on 29 October 2026 · Reviewed on 5 November 2026

Oil is pricing scarcity now; bonds are pricing inflation later

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    The primary indicator is whether the US 10-year yield stays above 5% through the ECB decision and the 3 November election. If it does, a December hike becomes all but certain; if it falls durably below 5%, the steepening of 28 October was a one-day reaction.

    What happened

    The yield did not stay above 5%. On the constant maturity series the 10-year was 4.94% on 30 October, 4.91% on 2 November and 4.95% on 3 November. The probability of a December hike also fell from 85% to about 76%. But the curve did not flatten: the 2s10s spread widened from 32 basis points on 28 October to 34 on 3 November. The level was a one-day reaction; the slope proved durable.

    Evidence: 3 November 2026 · Treasury yields climb on election day ahead of the refunding statement
  • Watched indicator

    Partly borne out

    The published claim

    The secondary indicator is the January–February Brent spread after the 30 October expiry: above $3, scarcity persists at the front of the curve; below $1, the December squeeze was an expiry effect.

    What happened

    The spread landed between the two thresholds, right on the upper one. According to MUFG, the spread between the first two contracts narrowed from $4 on 30 October to $3 on 2 November. It did not fall below $1, so scarcity has not left the front of the curve. But the condition of staying above $3 was borderline on 2 November; we found no sourced spread data for 3 and 4 November.

    Evidence: 2 November 2026 · Iran's Hormuz plan knocks 7.6% off January Brent in a day
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — A squeeze at both ends: the US 10-year yield stays above 5%, the January–February Brent spread stays above $3, and talks with Iran remain inconclusive but open.

    What happened

    Of the three triggers, one occurred in full, one was borderline and one did not occur. Talks stayed inconclusive but open: Qatar said on 3 November that it expected a reply to a new proposal within 36 hours, and Tehran did not drop its 7 conditions. The spread fell to $3 on 2 November. The yield, however, stayed below 5%, at 4.95% on 3 November.

    Evidence: 3 November 2026 · Qatari mediators give Washington and Tehran 36 hours to answer
  • Scenario

    Still open

    The published claim

    H2 · 30% — A round of strikes: the US and Israel hit Iran's energy and nuclear facilities; Iranian retaliation reaches Gulf refineries or merchant ships in Hormuz.

    What happened

    We set no deadline for this scenario, so the verdict remains open. As of 4 November there has been no strike on energy or nuclear facilities. On 2 November the US hit only a single missile launcher on Qeshm Island and said this was not a resumption of major operations. Trump said before the election that there would be no strikes, and left the option open after it.

    Evidence: 2 November 2026 · US hits launcher on Qeshm Island but says major campaign not resumed
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — Diplomatic thaw: Qatari mediation advances with a counter-proposal, NAVCENT eases its warning and the US 10-year yield falls durably below 5%.

    What happened

    The market side came out close to H3, the physical side close to H1. The yield fell below 5% and Qatar carried a new proposal. January Brent fell $8.66 to $104.77 on 2 November and closed at $103.60 on 3 November, its lowest since 7 October. But we found no source showing that the NAVCENT warning had been eased, and another tanker was hit off Oman on 3 November.

    Evidence: 2 November 2026 · Iran's Hormuz plan knocks 7.6% off January Brent in a day
  • Falsifier

    Partly borne out

    The published claim

    Suppose the US 10-year yield falls below 5% and the January–February Brent spread narrows below $1 after the expiry. Then the reading that scarcity is building at the front of the curve and cost at the far end weakens.

    What happened

    Half of the falsifier materialised. The yield has been below 5% since 30 October, at 4.91% on 2 November. The spread, however, did not fall below $1 and stood at $3 on 2 November. The reading has not been disproved but it has weakened: the far end retreated in level and stayed wide only in slope.

    Evidence: 3 November 2026 · Treasury yields climb on election day ahead of the refunding statement
  • Falsifier

    Still open

    The published claim

    Suppose the FEC and the Belgian finance ministry restore services within 7 days and the investigation finds the attack was random scanning. If no other election or tax portal goes down by 3 November, the reading that the attack targeted the final-day calendar weakens.

    What happened

    The seven-day deadline falls on 5 November. As of 4 November we found no official announcement that the FEC's electronic filing system or MyMinfin had been restored. The FEC home page showed no outage or reopening notice on the night of 5 November; that alone does not mean the system was back up. We also have no record of any other portal going down.

  • Watched indicator

    Confirmed

    The published claim

    Calendar item: according to Reuters sources, December production targets will not change at the 1 November OPEC+ meeting.

    What happened

    The 7 core OPEC+ members left their December targets unchanged on 1 November; the combined target stayed at roughly 31 million barrels a day. According to the IEA, Saudi Arabia produced 2.18 million barrels a day below its target in September, so the quota is not binding.

    Evidence: 1 November 2026 · OPEC+ freezes December quotas with effective spare capacity at 270,000 barrels

What we learned

We overstated the level in our bond reading but got the slope right. We tied the thesis that the far end was pricing inflation risk to a 5% yield threshold. A day later the yield fell to 4.98%, then to 4.91% on 2 November, and the probability of a December hike dropped from 85% to 76%. Even so, the 2s10s spread widened from 32 to 34 basis points, so the steepening continued. In oil it was a diplomatic headline, not physical data, that pulled the price down: the spread narrowed only from $4 to $3. Method: we will tie bond theses to level and slope separately rather than to a single level threshold. Because we set no deadlines for scenarios, H2 remained open; from now on every scenario will carry a date.

Published on 30 October 2026 · Reviewed on 5 November 2026

As insurers retreat, war risk moves onto state and corporate balance sheets

Open the brief ›
  • Watched indicator

    Still open

    The published claim

    The primary indicator is the first independently verified damage to the state-linked fleet in the southern Hormuz corridor. If verified, the last carrier filling the space insurance has vacated comes under risk too; if not, the mine claim remains a tool for raising costs.

    What happened

    Damage in the southern corridor was verified, but it was not verified that the damaged ship belonged to the state-linked fleet. UKMTO reported on 29 October that the Panama-flagged Louise 1 had been hit by a limpet mine off Hasab in Oman; the ship is linked to the Russian shadow fleet. On 3 November a tanker was hit by 2 projectiles 14 nautical miles off Lima; its owner was not disclosed. The Revolutionary Guards' supertanker claim could not be verified.

    Evidence: 30 October 2026 · First successful limpet attack of the war strikes tanker Louise 1 in Hormuz
  • Watched indicator

    Partly borne out

    The published claim

    The secondary indicator is the January–February Brent spread after the 30 October expiry: if it stays above $3, scarcity persists at the front of the curve.

    What happened

    According to MUFG, the spread between the first two contracts narrowed from $4 on 30 October to $3 on 2 November. Scarcity has not left the front of the curve, but the condition of staying above the threshold was borderline on 2 November. We found no sourced spread data for 3 and 4 November.

    Evidence: 2 November 2026 · Iran's Hormuz plan knocks 7.6% off January Brent in a day
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Balance sheets carry the load, supply continues: the state-linked fleet keeps loading without verified damage, the January–February spread stays above $3, and risk accumulates on state and corporate balance sheets.

    What happened

    The mechanism held, while the spread fell to the threshold. The UAE said on 2 November that its pipeline bypassing Hormuz had at one point lifted exports above pre-war levels; it gave no date or volume, so this part could not be verified. As of 2 November only one or two insurers were considering war-related event cancellation cover in the Gulf. The spread, meanwhile, fell to $3 on 2 November.

    Evidence: 2 November 2026 · War cover for Gulf event cancellation shrinks to a single insurer
  • Scenario

    Still open

    The published claim

    H2 · 30% — The last carrier is hit too: independently verified mine or attack damage to a state fleet tanker in the southern corridor; scarcity spreads across the whole curve.

    What happened

    We set no deadline for this scenario. As of 4 November there is no damage verified as belonging to the state fleet. Price moved the other way: January Brent closed at $103.60 on 3 November, its lowest close since 7 October.

    Evidence: 3 November 2026 · Qatari mediators give Washington and Tehran 36 hours to answer
  • Scenario

    Still open

    The published claim

    H3 · 15% — The cover door opens: Iran's 7-day plan is tied to a ceasefire and the marine war premium in Hormuz begins to fall back towards pre-war levels.

    What happened

    The trigger did not occur, but with no deadline set the verdict remains open. Araghchi said the Qatari plan could reopen Hormuz within 2 days, but the plan depends on US blockade and sanctions conditions and was not accepted. The Hormuz war premium stayed at 6–10% of hull value in October. Even so, the front of the Brent curve fell 7.63% on 2 November; without the trigger, price moved close to H3's market outcome.

    Evidence: 2 November 2026 · Iran's Hormuz plan knocks 7.6% off January Brent in a day
  • Falsifier

    Still open

    The published claim

    If a major exchange reopens service to Russian users by the end of November, or Yandex restores its guidance unchanged, the reading that digital payment and data infrastructure around Russia is narrowing weakens.

    What happened

    The deadline falls on 30 November. As of 4 November we have no sourced record of KuCoin or any other major exchange reopening service to Russian users. There is new data pointing towards narrowing: on 2 November Putin had 10% of private data centre capacity set aside for the state.

    Evidence: 2 November 2026 · Putin makes private data centres reserve 10% of capacity for the state
  • Falsifier

    Still open

    The published claim

    If war and cyber cover narrows at the January renewal while overall prices fall, the transfer of risk onto balance sheets accelerates; the cover threshold is what to watch.

    What happened

    The deadline is the 1 January 2027 renewal. Early signs point the same way as the thesis. SCOR's first January quotes are pulling European catastrophe pricing down, while as of 2 November only one or two insurers were considering war-related cancellation cover in the Gulf. No verdict is possible before the January results.

    Evidence: 30 October 2026 · SCOR's first January quotes point to lower European catastrophe pricing again
  • Watched indicator

    Confirmed

    The published claim

    Calendar item: no change to December production targets is expected at the 1 November OPEC+ meeting; the real question is the capacity of the fleet carrying Gulf exports.

    What happened

    The 7 core OPEC+ members left their December targets unchanged on 1 November; the combined target stayed at roughly 31 million barrels a day. Saudi Arabia produced 2.18 million barrels a day below its target in September; the binding constraint is not the quota but transport and capacity.

    Evidence: 1 November 2026 · OPEC+ freezes December quotas with effective spare capacity at 270,000 barrels

What we learned

We got the direction right on cover but again fell short on price. We correctly saw insurance withdrawing from the war layer; on 2 November only one or two insurers were pricing cancellation cover in the Gulf. But none of the 3 scenarios anticipated that a diplomatic headline alone could pull the price down; January Brent fell 7.63% on 2 November and dropped to $103.60 on 3 November. Our primary indicator also proved unmeasurable, since sources do not say which tanker belongs to the state-linked fleet. Method: when choosing indicators, we will check whether ownership or volume can be verified with public data, and we will set deadlines for scenarios.

Published on 27 October 2026 · Reviewed on 4 November 2026

Fiscal anchors, not geography, now draw the debt hierarchy

Open the brief ›
  • Falsifier

    Partly borne out

    The published claim

    If Mexican, Chilean and Italian premiums move by a similar magnitude to Brazil's on the next risk day, there is no divergence; if the gilt curve converges on the OBR's 4.5% assumption after the 28 October budget, the developed-world leg closes.

    What happened

    Divergence held over the week as a whole but broke down on a single day. Between 26 October and 2 November, Brazil's 5-year CDS rose from 137.43 to 145.32 basis points, while Mexico's fell from 97.34 to 95.51 and Italy's from 72.08 to 70.35. On the 30 October risk day, however, Italy rose from 69.12 to 74.11 and Brazil from 134.87 to 141.02, moves of similar size. Mexico rose by only 0.51 basis points that day. The gilt curve did not converge on the OBR assumption: the 10-year yield was 5.24% on 3 November. We found no dated data for Chile.

    Evidence: Investing.com — Brazil CDS 5 Years USD Historical Data
  • Watched indicator

    Still open

    The published claim

    If Brasília announces a rule tying the primary surplus to a number in early November, Brazil's 137.43 basis point CDS converges towards 120; if it delays, Argentina's 765 basis point country risk tests 800.

    What happened

    The deadline is early November, and as of 4 November we could not confirm from a source whether a rule had been announced. Interim data point the opposite way to convergence: Brazil's CDS stood at 135.70 basis points on 28 October, 141.02 on 30 October and 145.32 on 2 November, moving away from 120. We found no dated, citable source for Argentina's November country risk reading, so the 800 test could not be verified.

  • Watched indicator

    Partly borne out

    The published claim

    If the 10-year gilt yield rises above 5.36% after the 28 October budget, the developed-world leg of the divergence opens too; if it converges on the OBR's 4.5% assumption, it closes.

    What happened

    Neither threshold was triggered. The 10-year gilt yield was 5.24% on 3 November: below 5.36%, but 74 basis points above 4.5%. The yield has eased by about 15 basis points over the past month, and Danske Bank judged the budget cautious and in line with expectations. There was no budget shock; our reading that the developed-world leg was open found no confirmation through this channel, but the buffer problem has not closed either.

    Evidence: 3 November 2026 · Markets fully price a BoE hike, shifting debate to February
  • Watched indicator

    Confirmed

    The published claim

    The Bank of Canada's 28 October decision; 56% of the 34 respondents in the survey expect the rate to rise from 2.25% to 2.50%.

    What happened

    On 28 October the Bank of Canada raised its target for the overnight rate by 25 basis points to 2.50%, taking the bank rate to 2.75% and the deposit rate to 2.45%. The bank wrote that inflation was higher and more persistent than in its July report, and that it may need to raise rates again. The next decision is on 9 December.

    Evidence: Bank of Canada — Interest rate announcement, 28 October 2026
  • Scenario

    Still open

    The published claim

    H1 · 50% — Divergence deepens: Brasília announces no numerical spending rule in November but panic does not spread; Canada hikes on 28 October, the BOJ holds, and the UK budget shrinks the buffer.

    What happened

    The deadline falls at the end of November. Interim data are broadly consistent with H1: Canada raised rates to 2.50% on 28 October, and the BOJ held at 1.25% on 30 October by an 8–1 vote. Brazil's CDS rose to 145.32 on 2 November while Mexico's stayed at 95.51. We found no source with figures showing that the UK budget shrank the buffer; that condition could not be verified.

    Evidence: 30 October 2026 · Bank of Japan says it has hit its price goal, opening the door to December
  • Scenario

    Still open

    The published claim

    H2 · 25% — The common factor returns: the US 10-year exceeds 5.25%, or the budget reaction triggers a global sell-off at the far end of the gilt curve; the distinction between countries closes.

    What happened

    The trigger is distant for now. The US 10-year yield was 4.93% on 30 October, close to a 7-week low; the 10-year gilt was 5.24% on 3 November, with no post-budget sell-off. As the scenario runs to the end of November, there is no verdict yet.

    Evidence: 30 October 2026 · Logan calls for at least two more hikes as December odds slip to 76%

What we learned

The divergence thesis held over the week as a whole: Brazil's premium widened by 7.89 basis points while Mexico and Italy narrowed. But we wrote our falsifier around a single day, and on that day, 30 October, Italy widened by 5 basis points alongside Brazil. The common factor did not disappear; it simply worked selectively. We treated the gilt leg as evidence of divergence in the developed world; the budget passed quietly and this channel gave no confirmation. Method: we will write falsifiers over a window of at least 3 sessions rather than a single day, and we will not use thinly sourced gauges such as Argentina's as thresholds.

Published on 28 October 2026 · Reviewed on 4 November 2026

The oil bottleneck has moved from the wellhead to berths and refineries

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    The primary gauge is Ras Tanura's October loadings: if Kpler or Vortexa show an average above 3 million barrels a day, the 20 October picture proves temporary; if it falls below 2 million, the carrier of the Saudi recovery has broken.

    What happened

    The data landed between the two thresholds. According to Kpler, Ras Tanura loaded 2.95 million barrels a day over 1–25 October, down from 4.16 million in September, a fall of about 29%. The thinning was confirmed, but loadings did not drop to the 2 million breaking point. Among trackers, Kpler put total Saudi exports at 5.4 million barrels a day, Vortexa at 6.5 million and TankerTrackers at 6.99 million; the measurement gap makes it hard to rely on any single reading.

    Evidence: 31 October 2026 · Trackers disagree on Saudi exports by 1.6 million barrels a day
  • Watched indicator

    Partly borne out

    The published claim

    The secondary gauges are the number of dissents in the Fed's 28 October decision (possibly as many as 3) and whether the ECB revises its €75/MWh energy assumption on 29 October.

    What happened

    On 28 October the Fed held at 3.75–4.00% by a 10–2 vote; Cook and Hammack dissented in favour of a hike, making it 2 dissents rather than 3. The ECB held unanimously on 29 October, leaving the deposit rate at 2.50%. We found no wording indicating that the €75/MWh assumption had been revised; TTF, however, exceeded 100 euros on 30 October and stayed above the assumption.

    Evidence: 30 October 2026 · Logan calls for at least two more hikes as December odds slip to 76%
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Quiet thinning: the Ras Tanura fire stays outside the facility, October loadings come in below September; the Fed and ECB hold; Brent stays in a 110–120 dollar band.

    What happened

    The mechanism held; the price band did not. Ras Tanura loadings came in below September, the Fed and ECB held, and US distillate stocks fell to 10% below their 5-year average. But Brent broke out of the band: after the December contract expired on 30 October, January ICE Brent fell 8.66 dollars on 2 November to close at 104.77 dollars, recovering to 105.12 on 3 November. We had not factored the roughly 4-dollar spread at the contract roll into the band.

    Evidence: 2 November 2026 · Iran's Hormuz plan knocks 7.6% off January Brent in a day
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — Flows recover: Kpler or Vortexa show October loadings above 3 million barrels a day and the Qatari-mediated draft is accepted; Brent falls below 105 dollars.

    What happened

    The price outcome arrived through the wrong trigger. January Brent fell below 105 at 104.77 dollars on 2 November, but neither trigger materialised. Ras Tanura loadings stayed at 2.95 million barrels a day, and on 28 October Iran deemed the Qatari draft unacceptable. The fall was triggered by Araghchi saying the Qatari plan could reopen Hormuz within 2 days; the plan depends on US sanctions and blockade conditions and has not yet been accepted.

    Evidence: 28 October 2026 · Iran calls Qatar's draft unacceptable as Araghchi keeps the door open
  • Falsifier

    Still open

    The published claim

    If Bangladesh's 5-member committee attributes the outage to a one-off operator error, Congo's ARPCE reports in the first week of November that the disruptions have ended, and new breach notifications stop in Korea, the thin-layer reading weakens.

    What happened

    The deadline is the first week of November, that is, 7 November. As of 4 November we could not find sourced versions of the Bangladesh committee's report or ARPCE's November statement. Nor do we have a published record of a new breach at a Korean financial institution after 28 October; that does not mean notifications have stopped.

What we learned

We were right on the physical reading and wrong on the price reading. We correctly saw Ras Tanura thinning: Kpler shows 2.95 million barrels a day in October against 4.16 million in September. But we wrote a 110–120 band for Brent, and the January contract fell to 104.77 on 2 November. There were two errors. We did not factor the roughly 4-dollar spread at the contract roll into the band, and in H3 we tied the possibility of a diplomatic headline cutting the price by 8.66 dollars, without any change in physical data, solely to acceptance of the Qatari draft. The chance of 3 Fed dissents was also overstated; there were 2. Method: we will write price bands with the contract name and account for the roll; we will model diplomatic rhetoric as a trigger separate from acceptance.

Published on 23 October 2026 · Reviewed on 30 October 2026

With no barrels left to lose, Iran pushes oil risk onto neighbours

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    We called it noise: Moody's decision on 23 October would not be a breaking point for France, because most economists expected a one-notch cut from Aa3 to A1 and the move was largely priced in.

    What happened

    The conclusion held but the reasoning did not. Moody's made no cut on 23 October; the rating stayed at Aa3 with a negative outlook. The downgrade we treated as priced in never came. Nor was there a break: the OAT-Bund spread was 174.6 basis points on 22 October and 171.8 on 23 October, and on 27 October it narrowed 15.4 basis points in one session to 156.4.

    Evidence: 23 October 2026 · Moody's defies expectations and holds France's rating, outlook still negative
  • Falsifier

    Still open

    The published claim

    If by 6 November the France–Germany spread falls below 160 basis points, net reserves excluding swaps rise for two straight weeks to above 42 billion dollars and the semiconductor index recovers its 22 October loss, the collateral-scarcity reading weakens.

    What happened

    The deadline is 6 November. One of the three conditions has already been met in interim data: the France–Germany spread fell below 160, to 156.4 basis points, on 27 October. As of 30 October we could not find a sourced figure for net reserves excluding swaps after the week of 16 October. The falsifier requires all three conditions together, so there is no verdict yet.

    Evidence: 27 October 2026 · Italy's risk premium drops to 112 basis points, undercutting France
  • Scenario

    Still open

    The published claim

    H1 · 50% — The asymmetry persists and risk shifts to the neighbours: the blockade holds, no loadings resume at Kharg and Brent stays in a 105–118 dollar band.

    What happened

    No horizon was set for the scenario, and the H2 trigger depends on events after 3 November, so a verdict would be premature. Interim data fit H1: the ICE Brent December contract settled at 116.72 dollars on 23 October, 113.88 on 27 October and 114.88 on 29 October, inside the 105–118 band. Türkiye's CDS, however, crossed H2's 260 threshold without the trigger: 260.35 basis points on 23 October and 260.68 on 26 October.

    Evidence: Investing.com — Brent Oil Futures Contracts
  • Watched indicator

    Still open

    The published claim

    Primary indicator was Yanbu and Kharg: if Yanbu exceeds 4 million barrels a day by mid-November and Houthi attacks cause no Saudi output loss for 2 weeks, the risk has not shifted to the neighbours.

    What happened

    The deadline falls in mid-November. Attacks continued in the interim: on 24 October the coalition shot down 2 ballistic missiles fired at Yanbu and Taif. On 26 October the Houthis said they had struck pumping stations on the East–West pipeline; Saudi Arabia did not confirm it. We found no new measurement of Yanbu loadings beyond October's 3 million barrels a day.

    Evidence: 24 October 2026 · Coalition shoots down two Houthi missiles aimed at Yanbu and Taif

What we learned

We were right to call Moody's decision noise, but for the wrong reason. We wrote that the decision would be a priced-in downgrade; Moody's did not cut at all, and by 27 October the spread had narrowed from 171.8 to 156.4 basis points. The decision mattered little not because it was priced in, but because the channel carrying the premium was the budget calendar. We tied Türkiye's CDS only to the strike scenario, H2; on 23 October the CDS crossed the 260 threshold without any strike. Method: the reasoning behind a noise label will now be audited too, and when a threshold is tied to a scenario, the chance of crossing it without the trigger will be written down separately.

Published on 22 October 2026 · Reviewed on 29 October 2026

Energy shock demands fiscal support as bond markets charge a premium

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    Primary indicator was France's 10-year yield and Moody's decision on 23 October: if a one-notch downgrade left the spread settled above 160 basis points, the reading that budgets seeking energy support are punished with a debt premium would be confirmed.

    What happened

    The trigger never came: on 23 October Moody's kept France at Aa3 with a negative outlook. The spread widened without waiting for a downgrade: the OAT-Bund gap was 174.6 basis points on 22 October and 171.8 on 23 October. Once the rating held, the spread did not settle; on 27 October it narrowed 15.4 basis points in one session to 156.4. The debt-premium reading pointed the right way, but the channel carrying the premium was the budget calendar, not the rating decision.

    Evidence: 23 October 2026 · Moody's defies expectations and holds France's rating, outlook still negative
  • Watched indicator

    Falsified

    The published claim

    Secondary indicator was the MPC decision at 14:00 on 22 October: a deviation from 100 basis points or a change in forward guidance would set the direction of Turkey's CDS at 252.89 basis points.

    What happened

    There was no deviation: on 22 October the MPC cut by the expected 100 basis points, from 37% to 36%. The CDS found a direction anyway and rose: the Investing.com series shows 253.74 on 22 October, 260.35 on 23 October, 260.68 on 26 October and 258.30 basis points on 27 October. An MPC surprise was not what set the CDS direction; the indicator was of no use for this question.

    Evidence: 22 October 2026 · CBRT cuts rates for the first time in nine months as reserves fall
  • Scenario

    Still open

    The published claim

    H1 · 50% — The fiscal premium stays contained: Moody's holds the rating, the UK keeps energy support narrow in its 28 October budget, Brent stays in a $105–112 band, and the CBRT's cut keeps the CDS around 250.

    What happened

    The scenario matures with the BoE decision on 5 November. Interim data are mixed: Moody's held the rating on 23 October, but Brent broke above the band. The ICE Brent December contract closed at $116.72 on 23 October, $113.88 on 27 October and $115.32 on 28 October. The CDS did not stay around 250 either; it crossed H2's 260 threshold at 260.35 on 23 October and eased to 258.30 on 27 October.

    Evidence: 23 October 2026 · Moody's defies expectations and holds France's rating, outlook still negative

What we learned

Both of our indicators put the trigger in the wrong place. We tied France's premium to a Moody's downgrade; the rating held, while the spread widened to 174.6 before the decision and then fell to 156.4. We tied Turkey's CDS to an MPC surprise; the decision matched expectations and the CDS rose by more than 10 basis points. We repeated the lesson of the 21 October audit just one day later. Method change: the indicator we watch will now be a continuously measured variable of the channel we claim, not a single decision day.

Published on 21 October 2026 · Reviewed on 28 October 2026

The back end of the curve tightens while central banks wait

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    The primary indicator was the US 30-year yield: above 5.6%, housing, the French spread and emerging-market premia widen together; if the Freddie Mac rate on 22 October did not fall below 7.40%, the drop in contracts would carry into November.

    What happened

    The 30-year yield never approached the threshold: the FRED series shows 5.42% on 21 October, 5.29% on 22 October and 5.33% on 23 October. Freddie Mac cut the 30-year fixed rate from 7.40% to 7.32% in the week of 22 October. Yet the France–Germany spread hit a one-year high of 174.6 basis points on 22 October, even as the 30-year yield fell. The channel linking the French spread to the US 30-year yield did not hold; the spread widened on domestic budget risk.

    Evidence: 27 October 2026 · Italy's risk premium drops to 112 basis points, undercutting France
  • Watched indicator

    Partly borne out

    The published claim

    The secondary indicator was the MPC decision on 22 October: a deviation from the expected 36% would set the direction of Türkiye's CDS, then at 250.08 basis points.

    What happened

    There was no deviation: on 22 October the MPC cut the policy rate from 37% to 36%, in line with expectations. The CDS rose anyway, while we had expected an indecisive signal: the Investing.com series shows 253.74 basis points on 22 October, 260.35 on 23 October and 260.68 on 26 October. A decision that matched expectations did not anchor the CDS, and no MPC surprise set its direction. Net reserves also fell the same day, but no source attributes the CDS rise to a single cause.

    Evidence: 22 October 2026 · CBRT cuts rates for the first time in nine months as reserves fall
  • Scenario

    Still open

    The published claim

    H1 · 50% — The long end absorbs the tightening: the Fed holds on 28 October and the 30-year stays below 5.6%; the CBRT cuts by 100 basis points but external funding is capped by a 250 bp CDS, and the periphery spread compresses to around 150 bp.

    What happened

    The scenario falls due with the Fed decision at 21:00 on 28 October, after this review. Interim data are mixed. The CBRT cut by 100 basis points on 22 October and the 30-year stayed below 5.6%. But the CDS did not stay capped at 250 bp; it rose to 260.68 on 26 October. The French spread did not compress to around 150; it widened to 174.6 on 22 October before easing to 156.4 on 26 October.

    Evidence: 22 October 2026 · CBRT cuts rates for the first time in nine months as reserves fall
  • Causal chain

    Partly borne out

    The published claim

    The Russian oil premium flows to refiners; holding the rate at 14% on 23 October would be a response to budget risk more than to inflation.

    What happened

    The rate call was right: the Bank of Russia held its rate on 23 October and shifted its 2027 rate range up by 2 points. But the rationale was inflation, not the budget as we had written. The bank said underlying inflation had not fallen from an annualised 5–6% band, and that seasonally adjusted price growth rose to 8.0% in the third quarter. There is no evidence that the decision cited budget risk as its rationale.

    Evidence: 23 October 2026 · Bank of Russia holds rates and lifts its path for next year

What we learned

Both of our indicators weakened our reading before they ever reached their thresholds. We tied the French spread to the long end of the US curve and Türkiye's CDS to an MPC surprise; both moved independently of those links. The French spread peaked while the 30-year fell, and the CDS rose 10 basis points despite a decision that matched expectations. We also attributed the Russian rate decision to the budget; the bank said inflation. Lesson: pick the indicator from the local driver of the channel we claim, and never make a global variable the sole trigger for a country premium.

Published on 23 October 2026 · Reviewed on 28 October 2026

France trades like the periphery, and its marginal buyer sits outside Europe

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 55% — High premium, limited contagion: Moody's cuts by one notch or holds, the ECB holds on 29 October; the French spread settles in a 150–180 basis point band, and Belgian and Italian spreads widen without returning to 2012 dynamics.

    What happened

    The scenario falls due with the ECB decision on 29 October. Interim data: Moody's affirmed its Aa3 rating on 23 October. The OAT–Bund spread stayed inside the band at 171.8 basis points on 23 October and 156.4 on 26 October. The Italian leg went the other way, however: the BTP–Bund spread narrowed from 122 to 112 basis points on 26 October, and Italy settled below France. France's 5-year CDS stood at 104.18 on 23 October and 96.25 on 26 October.

    Evidence: 27 October 2026 · Italy's risk premium drops to 112 basis points, undercutting France
  • Falsifier

    Still open

    The published claim

    If the French parliament passes the revenue package, Moody's holds the rating and the spread falls below 150 basis points, foreign buyers' demand for a premium recedes and the chain breaks at its first link.

    What happened

    Only one of the three conditions was met. Moody's held the rating on 23 October. The same day the National Assembly rejected the revenue section of the social security budget, and the text went to the Senate; debate on the main budget begins this week. The spread stood at 156.4 basis points on 26 October, above 150. The chain did not break, and the test has not fallen due because the budget vote is still under way.

    Evidence: 23 October 2026 · Moody's defies expectations and holds France's rating, outlook still negative
  • Watched indicator

    Partly borne out

    The published claim

    Flagged as noise: Moody's decision on 23 October would not be a breaking point for France, because a one-notch downgrade was largely priced in.

    What happened

    A downgrade was priced in, but the decision was not a downgrade: Moody's held the rating. In the first session after the decision, on 26 October, the OAT–Bund spread narrowed by 15.4 basis points, from 171.8 to 156.4. No source attributes the whole move to the decision. But the decision we dismissed as noise coincided with the largest single-day move in the spread the report was tracking.

    Evidence: 27 October 2026 · Italy's risk premium drops to 112 basis points, undercutting France

What we learned

The report mentioned an affirmed rating only as a scenario trigger and never weighed it in the noise analysis; the 'downgrade is priced in' logic assumed the decision could only go one way. When the rating was held, the spread narrowed 15 basis points in a single session. On contagion, we also had Italy moving in the same direction as France; within three business days Italy traded below France. Lesson: in rating decisions, assign a separate probability to the opposite of the expected outcome, and do not treat the periphery as a single bloc.

Published on 20 October 2026 · Reviewed on 27 October 2026

The shock has moved from the wellhead to refineries and ships

Open the brief ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Tightness stays in delivery. Crude flows recover but distillate stocks remain more than 10% below the five-year average. Brent trades in a 100–115 dollar band while product margins and freight stay high.

    What happened

    The price band broke within 3 trading days. The ICE Brent front-month (December) contract closed at 116.72 dollars on 23 October and 115.81 dollars on 26 October; on the day of the brief it was 108.79 dollars. Trading Economics shows the 26 October close at 115.81 dollars and the morning of 27 October at 114.71 dollars. Distillate stocks stood at 104.3 million barrels in the week to 16 October, and the distance to the five-year average was not recalculated on that figure. The tightness leg of the scenario is open; the price leg failed.

    Evidence: Trading Economics — Brent crude oil
  • Watched indicator

    Partly borne out

    The published claim

    The primary indicator is the EIA weekly distillate stock. A further draw would strengthen the refinery constraint thesis, while a build above 2 million barrels would show product tightness past its peak.

    What happened

    On 21 October the EIA reported a build of 1.68 million barrels for the week to 16 October, against an expected draw of 1.16 million barrels; stocks rose from 102.6 to 104.3 million barrels. The draw that would have strengthened the thesis never came. The build also stayed below the 2 million barrel peak threshold, so the reading fell between the two thresholds and against the thesis. Data for the week to 23 October arrive on 28 October.

    Evidence: 21 October 2026 · US crude output drops by 410,000 barrels a day in a week
  • Scenario

    Still open

    The published claim

    H3 · 15% — The second ring breaks too. A direct hit lands on Yanbu or on an east coast terminal, or the Hormuz blacklist turns into an effective barrier to transit. Brent then moves above the band.

    What happened

    The effect leg appeared, the trigger was not confirmed. Brent closed above 115 dollars on both 23 and 26 October. The coalition said on 24 October that it had downed 2 Houthi missiles fired at Yanbu and Taif. The Houthis claimed the same day to have hit the East–West pipeline and the Rabigh refinery; Riyadh did not confirm it. There is no evidence of a direct hit on a terminal.

    Evidence: 24 October 2026 · Houthis claim strikes on Saudi East–West pipeline and Rabigh refinery
  • Watched indicator

    Still open

    The published claim

    The secondary indicator is the fill rate of EU gas storage. Daily injection falling below 0.10 points would mean the 80% reference for 1 November is missed.

    What happened

    The deadline falls on 1 November. The interim data confirm the direction: EU storage was 74.62% full on 22 October, and in a note on 23 October ING wrote that the fill appeared to have peaked just below 75%. That implies the 80% reference will be missed by roughly 5 points.

    Evidence: 23 October 2026 · European gas hits its highest since 2022 as storage peaks early
  • Falsifier

    Still open

    The published claim

    Converging signals: a recovery in US distillate stocks towards the five-year average through November, together with the diesel margin falling below 60 dollars, would break the chain at its first link.

    What happened

    The deadline is the end of November. In the first week stocks rose by 1.68 million barrels to 104.3 million barrels, which does not count as a recovery on its own. Among the events we have published there is no dated diesel margin figure after 20 October.

What we learned

We set the price band too narrowly. A 100–115 band written at 108.79 dollars lasted 3 trading days in a market moving 3% a day. Tying the mechanism of a scenario and a price range into the same sentence made the scenario fragile even where the mechanism was right. The distillate stock we track also came in against the thesis on its first reading. The lesson: derive the width of the band from realised volatility over the past 20 days, and write the price leg of a scenario as a separate indicator to watch.

Published on 22 October 2026 · Reviewed on 27 October 2026

A rate cut is coming, but Ankara's budget prices slow disinflation

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — A measured cut on a slow path. The CBRT cuts by 100 basis points on 22 October and gives cautious guidance. The CDS holds in a 245–260 band and a further cut follows in December.

    What happened

    The decision leg of the trigger happened. On 22 October the MPC cut the policy rate from 37% to 36% and wrote that the size of each step would be set meeting by meeting, with caution. The CDS leg failed on day one: the 5-year CDS closed at 260.35 on 23 October, above the band. The December cut leg falls due on 10 December.

    Evidence: 22 October 2026 · CBRT cuts rates for the first time in nine months as reserves fall
  • Watched indicator

    Still open

    The published claim

    Threshold: Türkiye's 5-year CDS moving above 260 would show the risk premium settling in and the December cut being called into question.

    What happened

    The threshold was crossed a day after publication: the CDS stood at 253.74 basis points on 22 October and 260.35 on 23 October. A single close does not prove persistence, and the source does not yet carry the 26 October close. The consequence leg of the claim, the questioning of the December cut, will be tested at the MPC meeting of 10 December.

    Evidence: Investing.com — Turkey CDS 5 Years USD (historical data)
  • Scenario

    Still open

    The published claim

    H2 · 25% — The risk premium halts the cycle: the fund law is delayed, reserve losses continue and the CDS moves above 260; the CBRT pauses in December or trims the step.

    What happened

    The trigger is mixed. The fund liquidation law was not delayed; parliament passed it on the evening of 22 October. Net reserves excluding swaps fell by 4.9 billion dollars in the week to 16 October, down to 39.34 billion dollars, and the CDS passed 260 on 23 October. The banking index fell more than 6% intraday on 23 October. The outcome of the scenario now rests on the December decision.

    Evidence: 22 October 2026 · Turkish parliament passes fund wind-down law with 1 million lira cap
  • Watched indicator

    Still open

    The published claim

    Threshold: the 10-year lira bond yield moving above 34% would show the market giving up on pricing a cutting path and demanding a term premium instead.

    What happened

    The data are unclear, so no verdict is possible. Investing.com shows 26 October at 35.785, with an open of 32.96 and a high equal to the close, while Trading Economics shows 35.79; the previous close was 33.01. A jump of 278 basis points intraday with no news flow raises the suspicion of a series break tied to a change in the benchmark bond. The threshold was not treated as crossed without confirmation from this series.

  • Falsifier

    Still open

    The published claim

    Condition that breaks the chain: inflation falling faster than expected and the CDS dropping below its 12 October level of 243.28 would speed the cuts and break the chain at its second link.

    What happened

    The condition has not been met and is moving the other way: instead of falling to 243.28, the CDS rose to 260.35 on 23 October. October CPI is published on 3 November, so the inflation leg of the condition has not yet been tested.

What we learned

We set the CDS band of H1 too close to the spot price. On the day of the report the CDS was 253.74 and the 260 ceiling was 6 basis points away; the band was pierced the next day. Reading the decision correctly did not save the scenario, because we had tied the risk premium not to the decision but to the fund crisis and to reserves. The 10-year yield threshold then became unmeasurable because of a suspected change in the benchmark. The lesson: check the continuity of a series before setting a threshold on it, and leave room in the band for spot volatility.

Published on 20 October 2026 · Reviewed on 25 October 2026

Energy's second round shows up on the street, not in wages

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — Street and rating together: Moody's cuts the rating to A1 on 23 October, protests spread into another weekend, and the French spread rises above 150 basis points and spills over to Italy.

    What happened

    The outcome arrived, the trigger did not. Reuters says the France–Germany 10-year spread reached 173.58 basis points on 22 October, its widest since November 2011. The Italian spread widened 10.5 basis points to 123 the same day, and the Belgian spread rose to 111.28. Moody's took no rating action on 23 October and kept France at Aa3 with a negative outlook. The spread broke the 150 threshold without a downgrade, one day before the decision.

    Evidence: 22 October 2026 · France's debt premium spills into Belgium as spread hits 2011 high
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Watered-down consolidation: the budget passes with concessions and Moody's holds the rating; the OAT–Bund spread stays high in a 120–150 basis point band and the ECB keeps waiting.

    What happened

    The Moody's leg was confirmed: the rating stayed at Aa3 on 23 October, conditional on the agency's assumption that a deficit-reducing 2027 budget is adopted. The spread band was falsified. The spread was 157.5 on 21 October in the Ideal Investisseur series and 173.58 basis points on 22 October according to Reuters, so the 150 ceiling was breached two days running. The budget leg is open: on 23 October the National Assembly rejected the revenue section of the social security budget.

    Evidence: 23 October 2026 · Moody's defies expectations and holds France's rating, outlook still negative
  • Causal chain

    Falsified

    The published claim

    If Moody's downgrades, the French spread rises above 150 basis points and spills over into Italian bonds; rating agencies and the bond market price political capacity.

    What happened

    The links ran in reverse order. The spread rose to 157.5 basis points on 21 October and 173.58 on 22 October, spreading to Italy at 123 basis points, all before Moody's decision on 23 October. Moody's did not downgrade. The market priced political capacity itself without waiting for the rating agency; the sequence in which we treated the rating channel as the trigger did not hold.

    Evidence: 22 October 2026 · France's debt premium spills into Belgium as spread hits 2011 high
  • Watched indicator

    Still open

    The published claim

    Threshold: a fall in euro/dollar from 1.1230 on 19 October to below 1.10 would show markets carrying political risk in France and Italy into the euro itself.

    What happened

    The threshold had not been crossed as of 23 October. The ECB reference rate was 1.1234 on 21 October, 1.1205 on 22 October and 1.1193 on 23 October. While the spread touched 173 basis points, the euro slipped only 0.3%. For now the pressure stays in the sovereign bond premium and has not passed to the currency.

  • Watched indicator

    Still open

    The published claim

    Threshold: a sustained fall in December Brent futures below 100 dollars would close the gap between headline and core inflation and drain fuel from street pressure (H3 · 15%).

    What happened

    It moved the other way. In the Investing.com series Brent closed at 108.79 dollars on 20 October, 113.17 dollars on 22 October and 116.72 dollars on 23 October. The 23 October close was the highest since 29 April. H3's trigger has receded, but it has not expired.

  • Falsifier

    Still open

    The published claim

    Condition that breaks the chain: if December Brent futures fall durably below 100 dollars, or France passes its budget without concessions, the rating channel stops at the first step.

    What happened

    The condition had not been met as of 23 October. Brent stands at 116.72 dollars. In France the main budget debate starts next week; the revenue section of the social security budget was rejected in the Assembly on 23 October and has gone to the Senate.

What we learned

We read the direction right and the sequence wrong. The report forecast that the French debt premium would exceed 150 basis points and spread to Italy; that happened within 2 days. But we tied it to a Moody's downgrade on 23 October: H2 was triggered by the downgrade and H1 carried a 120–150 band ceiling. The spread rose to 157.5 two days before the decision and 173.58 one day before, and Moody's did not cut the rating. The market moved ahead of the rating agency. Lesson: we should write the bond spread as an independent indicator tracked against its own threshold, not as the result of a calendar event. A scenario's trigger and its outcome should not be locked together.

Published on 20 October 2026 · Reviewed on 25 October 2026

Hormuz becomes a selective filter as Iranian exports hit zero

Open the report ›
  • Watched indicator

    Partly borne out

    The published claim

    Threshold: December Brent futures moving from 109.55 dollars on 19 October into a run of closes above 110 dollars would show markets pricing a scenario in which the filter narrows onto neighbouring producers.

    What happened

    In the Investing.com series Brent closed at 108.79 dollars on 20 October and 108.82 dollars on 21 October. It then stayed above 110 dollars two days running, at 113.17 dollars on 22 October and 116.72 dollars on 23 October. Trading Economics also gives the 23 October close as 116.72 dollars. Two closes are too few for a run; the week of 26 October will decide.

    Evidence: Investing.com — Brent futures historical data
  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — The filter narrows onto neighbours: the list extends to tankers carrying Iraqi and Kuwaiti cargoes or the IRGC seizes a listed ship; Iraqi exports fall below 3 million barrels a day and Brent settles above 110 dollars.

    What happened

    The price leg arrived, the trigger was not confirmed. On 20–21 October a Kuwaiti tanker was hit off Iraq's Faw and a ship was fired on at the Basra anchorage. The perpetrator was not confirmed, and neither event was a list expansion or a seizure. Iranian state television (state media) announced on 23 October that transit by ships not following the rules had stopped. No dated Iraqi export figure after 19 October could be found.

    Evidence: 21 October 2026 · Kuwaiti tanker hit in Hormuz as October attacks double
  • Watched indicator

    Confirmed

    The published claim

    Noise: Kpler measuring Gulf exports at 8.5 million barrels a day in one week does not show that flows have collapsed.

    What happened

    Later measurements supported this reading. WSJ data cited by gCaptain put Hormuz flows, including ship-to-ship transfers at sea, at about 82% of pre-war levels as of 22 October. Reuters columnist Ron Bousso gave the same ratio, citing Kpler. The weekly reading of 8.5 million barrels did not on its own signal a collapse.

    Evidence: 22 October 2026 · Revolutionary Guards claim seven ships hit, but UKMTO finds no match
  • Watched indicator

    Confirmed

    The published claim

    Signal: Iran's crude exports have been at zero since late August, so restricting the strait now costs Tehran nothing.

    What happened

    A WSJ report based on Kpler says Iran's floating storage in Asia fell below 1 million barrels on 5 October and was exhausted by mid-October. On 22 October, the 100th day of the blockade, CENTCOM said at least 25 billion dollars of exports had been blocked. In the same days the Revolutionary Guards threatened to strike US bases, and Iranian state media announced that transit by non-compliant ships had stopped.

    Evidence: 22 October 2026 · Iran's floating oil stocks in Asia have run dry
  • Watched indicator

    Partly borne out

    The published claim

    Threshold: WTI rising from 103.66 dollars to above 105 dollars and the Brent spread narrowing below 5.9 dollars would show the Gulf gap being filled with US crude.

    What happened

    The site's wti indicator is the FRED spot series, and it was already above the threshold when the report was written: 110.01 dollars on 19 October and 107.89 dollars on 20 October. In the same series the Brent–WTI spot spread was 8.81 dollars on 19 October and 7.33 dollars on 20 October; it did not fall below 5.9 dollars. Half the threshold had been crossed at the time of the report, and the other half did not happen.

    Evidence: FRED — Crude Oil Prices: West Texas Intermediate (DCOILWTICO)
  • Watched indicator

    Still open

    The published claim

    Threshold: daily PortWatch Hormuz transits rising from 2 on 11 October to above 20 would show the ship-by-ship permit regime loosening and the list losing its function (H3 · 15%).

    What happened

    The PortWatch series stops at 18 October and the threshold was not crossed: daily transits from 14 to 18 October were 3, 2, 7, 0 and 4. UKMTO counts 114 exits and 127 entries over the past 7 days, but its method differs, so it cannot be compared with the PortWatch threshold.

  • Falsifier

    Still open

    The published claim

    Condition that breaks the chain: if the US suspends the blockade, Iran's own exports return; if escorted transits on PortWatch rise above 20 a day, the list becomes ineffective.

    What happened

    The condition had not been met as of 23 October. CENTCOM said the blockade entered its 100th day on 22 October, and the US Treasury sanctioned a law firm linked to Iran's shadow fleet the same day. PortWatch transits were 4 on 18 October.

What we learned

The Iran leg of the thesis held; we could not measure the Iraq leg. Iran's zero exports and exhausted floating storage were confirmed, and our call to treat the 8.5 million barrel Kpler reading as noise was backed by the later 82% figure. Brent passed 110 dollars, but the list expansion that was H2's trigger was not confirmed; prices reacted instead to attacks and Iranian threats. Method error: we wrote the WTI threshold against the futures contract, whereas the site's wti indicator is the FRED spot series and was already above the threshold when the report came out. When setting thresholds, we should check the latest value of the indicator series used on the site.

Published on 18 October 2026 · Reviewed on 24 October 2026

Crude is flowing again, but the fuel ships burn is not

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — The attack zone widens: strikes in the northern Gulf loading waters become weekly, or the loss of a supertanker is independently confirmed; December Brent settles above $110 and tanker freight spikes.

    What happened

    The trigger arrived. In the week after the 16 October Jubail attacks, a tanker was hit 30 nautical miles south-east of Faw on 20–21 October, and 4 projectiles were fired at a vessel transferring cargo at Basra. No supertanker loss was independently confirmed; UKMTO did not match the Revolutionary Guards' 22 October claim of 7 ships. Brent closed at $113.17 on 22 October and $116.72 on 23 October; two closes are still too few to call it durable.

    Evidence: 21 October 2026 · Kuwaiti tanker hit in Hormuz as October attacks double
  • Watched indicator

    Confirmed

    The published claim

    Signal: the attack zone has reached the loading waters of the northern Gulf; on 16 October 2 tankers were hit 68 and 78 nautical miles north-east of Jubail.

    What happened

    The shift north continued into a second week. On 20–21 October a tanker was hit off Iraq's Faw, and a vessel at the Basra anchorage came under fire. UKMTO recorded 32 attacks around the Arabian Peninsula in October, against 16 in the whole of September. The perpetrator was not confirmed in any incident.

    Evidence: 21 October 2026 · Kuwaiti tanker hit in Hormuz as October attacks double
  • Watched indicator

    Partly borne out

    The published claim

    Threshold: December Brent settling above $110 would show that prices, too, accept the attack zone has widened.

    What happened

    According to the Investing.com series, Brent closed at $109.55, $108.79 and $108.82 on 19–21 October. It then held above the threshold for two days running, at $113.17 on 22 October and $116.72 on 23 October. The 23 October close is the highest since 29 April. The run covers only 2 days; durability needs at least 1 week.

    Evidence: Investing.com — Brent Oil Futures Historical Data
  • Scenario

    Still open

    The published claim

    H1 · 50% — Crude recovers, fuel stays tight: Hormuz crude flows hold at 65–80% of normal, the Gulf fuel oil shortfall persists at 600,000–1 million tonnes a month, and strikes on Russian refineries continue.

    What happened

    The crude leg got lost between measurements. Kpler put Gulf exports at 8.5 million barrels a day in the week before 19 October, down from more than 20 million the week before. Washington gave flows at 80% of pre-war levels, and WSJ data at 82%. Strikes on Russian refineries continued. No dated figure after 18 October could be found for the Gulf fuel oil shortfall.

  • Scenario

    Still open

    The published claim

    H3 · 20% — Substitution closes the gap: Asian refiners are fed with crude again, low-sulphur fuel from Dangote and China plus LNG bunkering meet part of demand, and the Singapore premium falls below $20 a tonne.

    What happened

    No public, dated value after 18 October could be found for the Singapore 380 cst premium. Asian refiners did not give up on crude, but freight costs pushed them to smaller tankers. Taiwan's CPC bought 6 Aframax cargoes in one week, and shipping US crude to Asia rose to $40 a barrel.

  • Causal chain

    Still open

    The published claim

    Shipowners pass rising fuel costs into freight through bunker adjustment surcharges; as attacks spread to the northern Gulf, a war-risk premium is added to the same cargo; dearer freight feeds into Türkiye's foreign trade bill.

    What happened

    The links cannot yet be measured. As of 23 October, container lines' November bunker surcharge announcements were not available, and TurkStat's October foreign trade data are due at the end of November. On the crude side freight is already at a peak; UTIKAD says it has reached $40 a barrel, but that stems from a shortage of ships, not fuel.

  • Falsifier

    Still open

    The published claim

    Condition that breaks the chain: Gulf fuel oil exports return above 3.5 million tonnes a month, Asian refiners are fed with crude again, or strikes on Russian refineries stop.

    What happened

    The condition had not been met as of 23 October. Strikes on Russian refineries have not stopped. According to Bloomberg, Ryazan has been offline for more than half the time since May, and 5 of the 15 confirmed strikes in September hit Rosneft facilities. There are no dated data for Gulf fuel oil exports for late October.

What we learned

We could measure the price leg of the scenarios but not the core thesis. The report argued that scarcity had shifted from crude to bunker fuel. Its indicator, the Singapore 380 cst premium, is not available from any free, dated source, so in 6 days we could not test the heart of the thesis. The northward shift in attacks and Brent above $110 support H2. Yet no data link Brent's 7% rise on 22–23 October to the fuel oil channel; Houthi attacks on Saudi facilities dominated those same days. The lesson: if the indicator at the centre of a report has no public, dated series, we should name a measurable proxy instead of making it a threshold.

Published on 18 October 2026 · Reviewed on 24 October 2026

Russian diesel ties Ankara's fuel security to a Washington–Moscow bargain

Open the report ›
  • Watched indicator

    Confirmed

    The published claim

    Signal: the source of Russian diesel is under attack; the CDU-6 unit at Ryazan stopped after the 15 October strike, the fourth on the plant this year.

    What happened

    According to Bloomberg's 22 October damage assessment, Ryazan has been offline for more than half the time since the first strike in May. Syzran, which processes about 9 million tonnes a year, was idle or ran below a third of capacity for most of 6 months. Of the 15 refinery strikes confirmed in September, 5 hit Rosneft facilities. The figures could not be independently verified.

    Evidence: 22 October 2026 · Rosneft starts to find the refinery repair bill unsustainable
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Diesel flows, the bill stays high: the US–Russia diesel deal and the Graham waiver hold, Ukrainian strikes stay intermittent; the Russian share stays low but stable and energy inflation rises modestly.

    What happened

    The trigger conditions were in place as of 23 October. In Moscow on 21 October, Witkoff thanked Putin on Trump's behalf for the diesel decision, and the waiver was not withdrawn. The strikes were intense rather than intermittent; damage to Rosneft facilities was documented on 22 October. The scenario's impact leg could not be measured: EMRA's October import breakdown and October CPI arrive in early November.

    Evidence: 21 October 2026 · Putin tells Witkoff he backs mid-November three-way talks in principle
  • Scenario

    Still open

    The published claim

    H2 · 30% — The Russian export link breaks: strikes spread to Baltic and Black Sea terminals, Russian diesel exports shrink markedly; energy jumps in October and November CPI and the CBRT slows its cuts.

    What happened

    There is a sign on the Baltic leg, but its cause is unclear. Shipments from Primorsk stopped for about 3 days, and Russian seaborne crude exports fell to 3.13 million barrels a day in the week to 18 October, the lowest since March. The halt was officially put down to technical work. There are no dated diesel export figures after 18 October. The CBRT cut by 100 basis points on 22 October, as expected; there was no slowdown at the first meeting.

  • Scenario

    Still open

    The published claim

    H3 · 20% — Washington withdraws the waiver: pressure from Congress or collapsing peace talks lead to a partial lifting of the waiver; Türkiye's purchases of Russian energy again face the risk of US tariffs.

    What happened

    The waiver remained in force as of 23 October. Talks have not collapsed: on 21 October Putin agreed in principle to a trilateral meeting in mid-November. That same night Russia struck Kyiv with missiles, and Lavrov said Moscow would make no political concessions. The 60-day review period Rubio asked for runs out in mid-December.

  • Causal chain

    Still open

    The published claim

    As expectations deteriorate, the CBRT keeps its easing path below the 265 basis points over 3 meetings priced in surveys; the energy bill pushes current-account deficit expectations higher.

    What happened

    The first data point runs against the chain. The MPC cut the policy rate by 100 basis points on 22 October, and 29 of 33 economists in the Matriks survey expect another cut on 10 December. The decision text still flagged energy prices as an upside risk. The 3-meeting window has not closed; October CPI is due on 3 November and the Inflation Report on 12 November.

  • Falsifier

    Still open

    The published claim

    Condition that breaks the chain: Kyiv and Moscow agree an energy ceasefire covering refineries and terminals, or a ceasefire on civilian shipping in the Black Sea takes hold.

    What happened

    The condition had not been met as of 23 October. No ceasefire covering energy infrastructure was announced, and Russia struck Kyiv hours after the Witkoff meeting. The trilateral talks have been put off until mid-November.

What we learned

The report's supply side holds; its monetary policy link failed its first test. The erosion at Russian refineries was documented 4 days later by Bloomberg's damage count. But the final link in the chain said energy-driven damage to expectations would slow the CBRT. On 22 October the committee delivered the expected 100 basis points and merely noted the energy risk in its statement. One meeting does not make an error, but the direction is against us. The lesson: when tying an inflation link that plays out over weeks to a rate decision due the same week, we should first have stated how much of that decision the market had already priced.

Published on 17 October 2026 · Reviewed on 23 October 2026

Central banks pause in October as energy persistence decides December

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 55% — Hold in October, synchronised hikes in December: euro-area services inflation stays above 3.2% and Brent holds above $100. The Fed and the ECB hold in October and raise by 25 basis points each in December, while the UK makes its first hike in November.

    What happened

    The decisions will come at the Fed meeting on 28 October and the ECB meeting on 29 October. The energy leg of the trigger is intact: Brent closed between $108.79 and $112.92 on 19–22 October. On 19 October Barclays said it expects the Fed to hike twice, in December and January; markets price a 16% probability for October and 92% for December. In the UK, September CPI rose to 3.5% and a November hike is priced in.

  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: the US 10-year yield returning above 5.36%; Brent settling durably above $110; the euro falling below 1.11 against the dollar.

    What happened

    Only the Brent threshold was crossed, and only once: it closed at $112.92 on 22 October after three previous closes below $109, so there is no persistence. The US 10-year yield stood at 5.15% on 19 October, 21 basis points below the threshold. The euro was at 1.1234 against the dollar on 21 October, above 1.11.

    Evidence: Investing.com — Brent futures historical data
  • Watched indicator

    Confirmed

    The published claim

    Noise: Bailey judged second-round effects to be weak and the UK will stay on hold. The real signal is that most economists expect a hike in November.

    What happened

    On 21 October the ONS put September CPI at 3.5%, against expectations of 3.4%; core rose from 2.6% to 2.8% and services inflation from 3.4% to 3.7%. The 0.3-point rise in services contradicts the reading that second-round effects remain weak. The decision is due on 5 November; our call to treat the on-hold reading as noise is holding for now.

    Evidence: 21 October 2026 · UK inflation jumps to 3.5%, all but sealing a November hike
  • Watched indicator

    Falsified

    The published claim

    Signal: Canada's room to look through energy is narrowing; on 16 October Macklem said high oil prices carry a broader inflation risk.

    What happened

    Canadian data showed the opposite. September inflation came in at 3.1%, below the 3.2% expected, and the rate excluding gasoline fell from 2.4% to 2.2%. Among core measures, CPI-trim eased from 1.9% to 1.8% and CPI-median from 2.0% to 1.9%. On the 19 October data, the energy shock is not spreading into services and core prices in Canada; the room to look through energy has been preserved, not narrowed.

    Evidence: 19 October 2026 · Canadian inflation rises to 3.1% while core gauges stay below target
  • Causal chain

    Still open

    The published claim

    The October meetings pass with a hold; if early-November data confirm services prices, the ECB and the Fed hike together in December. As the rate floor rises, the CBRT's room to cut narrows from the external side.

    What happened

    The first link will be tested by the 28–29 October decisions and the second by Eurostat's flash estimate on 4 November. The indicator for the third link, Türkiye's 5-year CDS, stood at 251.06 basis points on 21 October, against 248.07 on 16 October. We have not yet published a record on the CBRT's 22 October decision.

  • Falsifier

    Still open

    The published claim

    The condition that breaks the chain: if Brent falls below $100 and euro-area services inflation returns to 3.0%, the December hikes are postponed; if US employment deteriorates quickly, the Fed extends its hold.

    What happened

    As of 23 October the condition has not been met: Brent closed at $112.92 on 22 October, $12.92 above $100. Euro-area services inflation for October is due on 4 November. US employment data for October have not yet been released; only 29,000 jobs were added in September.

What we learned

We got one signal wrong. We cast Canada as a country whose room to look through energy was narrowing, yet in the September data released on 19 October both core measures fell and inflation excluding gasoline dropped to 2.2%. We took the governor's rhetoric in place of the data. The main scenario matures on 28–29 October, and the verdict will be given then. The lesson: before central bank rhetoric is counted as a signal, it will be checked against the same country's latest core inflation data.

Published on 17 October 2026 · Reviewed on 23 October 2026

Tehran puts a political price on Hormuz, and oil futures price scarcity

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Conditional passage persists: Iran keeps its coordination requirement, Strait of Hormuz flows stay in a band of 60–75% of normal and attacks on Yanbu are averted. Producers with weak alternative outlets, such as Iraq, keep losing revenue.

    What happened

    The coordination requirement held: Iran expanded its blacklist to 122 vessels on 19 October, and the Revolutionary Guards declared the strait fully closed on 21 October. The band leg could not be measured: Kpler put Gulf exports at 8.5 million barrels a day in the week before 19 October, while Washington described flows as 80% of pre-war levels. The Iraq leg proved wrong; exports rose above 3.4 million barrels a day in the first half of October, back to pre-war levels.

    Evidence: 19 October 2026 · Iraq's oil exports return to pre-war levels in October
  • Scenario

    Partly borne out

    The published claim

    H2 · 25% — Both outlets narrow at once: Iran steps up attacks on uncoordinated transits and the Yanbu terminal or its storage takes a hit; Brent futures settle durably above $110.

    What happened

    The attack leg arrived, but the perpetrator was not verified: UKMTO counted 32 attacks in October, against 16 in September, and a Kuwaiti tanker was struck while transiting the strait on 21 October. The Yanbu leg did not: the Houthis claimed on 19 October to have hit Yanbu and Shaybah, but according to Argus neither facility was damaged. Saudi exports nonetheless fell to 5.55 million barrels a day in the first half of October. Brent closed at $112.92 on 22 October; a single day does not count as persistence.

    Evidence: 21 October 2026 · Kuwaiti tanker hit in Hormuz as October attacks double
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — An interim arrangement takes shape: an Iran–Oman corridor comes into operation, or the US and Iran trade passage for an easing of the blockade and sanctions; Iraqi exports approach pre-war levels.

    What happened

    Half the trigger arrived: Araghchi said on 20 October that a special transit route had been opened for Iraqi oil and that coordinated routes had been agreed with Oman; the routes await IMO registration. There is no political deal with the US. The scenario's effect on Iraq did materialise, with exports above 3.4 million barrels a day in the first half of October. That figure has not yet been confirmed by independent ship tracking.

    Evidence: 20 October 2026 · Iran says it has opened dedicated Hormuz lanes for Iraqi oil
  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: Brent December futures moving into a run of closes above $110; daily Hormuz transits falling below 20; US commercial crude stocks falling below 410 million barrels.

    What happened

    Brent closed at $109.55, $108.79 and $108.82 on 19–21 October, and crossed the threshold for the first time at $112.92 on 22 October; there is no run yet. The Hormuz threshold had already been breached on the day of publication: PortWatch counted 7 transits on 16 October and 4 on 18 October. US crude stocks stood at 412.5 million barrels in the week of 16 October, above the threshold.

    Evidence: Investing.com — Brent futures historical data
  • Watched indicator

    Falsified

    The published claim

    Signal: losses are becoming entrenched for the Gulf producer with weak alternative outlets; Iraqi exports fell from 3.5 million to 2.6 million barrels a day, and foreign exchange reserves shrank by $20 billion.

    What happened

    The 2.6 million barrel figure we used covered only the first three weeks of September; according to Iraq's ministry, the average for the whole of September was about 3 million barrels. In the first half of October exports rose above 3.4 million barrels a day. With only 200,000–230,000 barrels a day moving through the Ceyhan pipeline, the recovery depends on Hormuz; the claim of persistence was reversed within 2 weeks.

    Evidence: 19 October 2026 · Iraq's oil exports return to pre-war levels in October
  • Causal chain

    Still open

    The published claim

    Because the political condition cannot be met, the gap is covered by reserve releases. The US reserve falls from 283 million barrels towards 266 million, and the shortage then passes into coal and the external balances of importing countries.

    What happened

    The first link has not yet engaged: according to the EIA, the strategic reserve slipped from 283.0 million to 282.7 million barrels in the week of 16 October, a fall of only 0.3 million barrels. The result of the 17 million barrel exchange tender has not been announced. Data for the coal link and for TÜİK's October foreign trade figures are due in November.

  • Falsifier

    Still open

    The published claim

    The condition that breaks the chain: the temporary Iran–Oman corridor is submitted to the IMO and enters use, or the US and Iran set up an interim arrangement trading passage for an easing of the blockade and sanctions.

    What happened

    As of 23 October the condition has not been met: Iran and Oman have agreed route coordinates, but IMO registration is pending. Araghchi said on 20 October that the technical arrangement would not open the strait to general traffic without a political agreement. The Revolutionary Guards declared the strait fully closed on 21 October.

What we learned

We were wrong in two places. First, we set the Hormuz threshold without checking it against our own series: PortWatch was counting 7 transits on the day of publication, so fewer than 20 a day was the starting point rather than a break, and the threshold measured nothing. Second, we treated Iraq's loss as lasting but used a 2.6 million barrel figure that covered only the first three weeks of September. In October exports returned to 3.4 million barrels and the Iraq leg of H1 proved wrong. The lesson: thresholds will be written against the latest value on the day of publication and the series from the same method, and no single-country figure will count as evidence of persistence until its period definition has been checked.

Published on 16 October 2026 · Reviewed on 22 October 2026

Treasuries are pricing AI debt, not inflation

Open the brief ›
  • Scenario

    Still open

    The published claim

    H1 · 50% — The haven holds and the Fed waits. High-yield spreads stay at 300–360 basis points and the 10-year yield stays below 5.33%. The Fed holds on 28 October; the CBRT cuts by 100 basis points on 22 October.

    What happened

    Not yet due: the Fed decides on 28 October and the MPC announces at 14:00 on 22 October. The interim picture favours H1. The 10-year yield rose to 5.144% on 19 October but stayed below 5.33%. Markets price a 16% chance of an October hike and roughly 92% for December.

    Evidence: 19 October 2026 · Barclays now expects Fed hikes in both December and January
  • Scenario

    Still open

    The published claim

    H2 · 30% — Inflation takes the lead again. The credit spread falls below 300 basis points and the odds of an October hike pass 50%. The Fed raises rates by 25 basis points on 28 October.

    What happened

    Not yet due. For now the trigger is receding: the odds of an October hike fell from 33% to about 15% on 16 October and stood at 16% on 19 October. Hike expectations have shifted from October to December. That weakens the trigger of H2, not its timing.

    Evidence: 16 October 2026 · Warsh gives no rate signal in Bangkok and October hike odds fall
  • Watched indicator

    Falsified

    The published claim

    We named a move in Türkiye's CDS above 245 basis points as the market footprint of H3 (20%, credit stress spilling into the Treasury market).

    What happened

    Türkiye's 5-year CDS rose to 250.08 basis points on 19 October and 252.89 on 20 October, so the threshold was crossed. Yet the H3 trigger never formed: the high-yield spread did not approach 400 basis points and there was no margin-call selling in Treasuries. A domestic channel carried the premium. The heads of the SPK and KGK were dismissed on 20 October, a day after the BIST 100 fell 2.49%. Treating the CDS as a footprint of the global credit scenario was wrong.

    Evidence: 20 October 2026 · Official Gazette removes the heads of Türkiye's market and audit regulators
  • Falsifier

    Still open

    The published claim

    Converging signals: the reading that 'safe zones are shrinking' fails on three conditions. By 30 November the Romanian inquiry rules out a drone link to Wine Trader and no new container ship is hit in Hormuz. The Bankruptcy Notification Centre also reopens within 2 weeks without data loss.

    What happened

    Not yet due: the 2-week window for the notification centre closes on 29 October and the deadline for the Romanian inquiry is 30 November. Events we published between 16 and 21 October show no new container ship hit in Hormuz. The attacks stayed on tankers, including 2 off Jubail on 16 October.

    Evidence: 16 October 2026 · Two tankers hit off Jubail as attacks reach the northern Gulf
  • Watched indicator

    Still open

    The published claim

    Foreign flows into government bonds in the first weekly securities data after the 22 October MPC decision; if the 451.1 million dollar inflow of the 9 October week reverses, the reserve recovery reverses too.

    What happened

    The MPC decision is due at 14:00 on 22 October; the first weekly data after the decision has not yet been published.

What we learned

The brief's main scenarios hinge on the Fed decision of 28 October. Ruling on them today would mean writing a forecast, so they stay open. We do have one clear miss. We cast a move in Türkiye's CDS above 245 basis points as the market footprint of global credit stress (H3). The threshold was crossed within 4 days, but without any H3 trigger; domestic channels such as the SPK–KGK dismissals and fund outflows carried the premium. The lesson: before using the Turkish premium as a tracker for a global scenario, model the domestic channel as a separate scenario variable. The probabilities stand as written.

Published on 16 October 2026 · Reviewed on 22 October 2026

Oil is flowing again, but the binding constraint is now stocks

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Flows hold, the buffer drains: Gulf crude exports hold at 13–15 million barrels a day. Remaining emergency stocks are released by year-end, and diesel stays high, decoupled from crude.

    What happened

    The buffer half held. US distillate stocks fell to 102.6 million barrels in the week of 9 October, the lowest for that week of the year since 1982, and the diesel crack reached $97.09 a barrel on 19 October. The flow half did not. Kpler measured Gulf crude exports at about 8.5 million barrels a day in the week before 19 October, far below the band. Saudi exports slipped to 5.55 million barrels a day over 1–15 October. The measurements conflict: on the same day the US Energy Secretary said flows had recovered to about 80% of pre-war levels.

    Evidence: 19 October 2026 · Kpler sees Gulf exports at 8.5 million barrels as Washington claims recovery
  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — The corridor breaks: attacks on transfer and waiting areas in the Gulf of Oman intensify or a mine threat is confirmed, and Strait of Hormuz transits fall below 20 a day.

    What happened

    Attacks did intensify, with at least 10 verified strikes on merchant ships in the week before 19 October. The mine threat was not confirmed. The Revolutionary Guards announced 3 mine claims in 3 days, and none was corroborated by an independent source. The transit threshold was poorly calibrated. IMF PortWatch put Hormuz transits at 7, 0 and 4 ships on 16–18 October, but the same series was already below 20 a day before the report was written (7 on 13 October).

    Evidence: 19 October 2026 · IRGC announces a third mine claim with no independent confirmation
  • Watched indicator

    Partly borne out

    The published claim

    Threshold: December Brent futures breaking above $110 from $105.92 on 15 October would show physical scarcity feeding into the futures curve.

    What happened

    Breached intraday, not on a close. Brent touched $111.99 on 20 October but settled at $108.79. Closes have held between $108.79 and $109.86 since 16 October, with $108.82 on 21 October. The physical–futures gap is narrowing, but through falling physical prices rather than rising futures: Dated Brent dropped to about $117 on 19 October, down roughly $16 in a week.

    Evidence: FinanceFeeds — Brent Oil Price Today: $110, Bull $113 vs Bear $105
  • Watched indicator

    Partly borne out

    The published claim

    Threshold: US commercial crude stocks falling below 410 million barrels from 415.6 million in the week of 10 October would show US inventories still covering the Gulf shortfall.

    What happened

    The draw continued but the threshold was not crossed. EIA data show commercial crude stocks at 412.5 million barrels in the week of 16 October, against 415.6 million a week earlier. In the same week distillate stocks rose from 102.6 to 104.3 million barrels. It is the first data point to contradict the reading of a product buffer draining in one direction only.

    Evidence: EIA — Weekly Petroleum Status Report, Highlights

What we learned

The buffer thesis largely held: distillate stocks hit a 42-year low and the diesel crack sits in record territory. We got two things wrong. First, we wrote that Gulf flows would hold in a 13–15 million barrel band; a week later Kpler measured 8.5 million, and the flow leg of our most likely scenario collapsed. Second, we set the H2 Hormuz threshold of 20 transits a day without checking it against the IMF PortWatch series we show on the site. That series was already below the threshold, so the threshold distinguished nothing. The lesson: a threshold is written on the scale of the indicator the reader sees, and checked against the latest value before publication. The probabilities stand as written.

Published on 22 September 2026 · Reviewed on 21 October 2026

Africa's fuel problem is no longer price but the last mile: the valve, the road and the pump are cut separately

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    S1 · 55% — Disruptions stay local: the valve in Libya reopens within days, the convoy system in Mali holds, the drone wave in Sudan stays confined to El Obeid, and Brent holds a 95–110 range.

    What happened

    The Libya and price legs held. The Sharara–Zawiya line reopened on 26 September, six days later, and Brent futures sat inside the range at 108.79 dollars on 20 October. The Sudan leg did not. Drone strikes spread well beyond El Obeid: on 16–19 October 41 people were killed in Blue Nile, 17 in Nyala and 4 in Khartoum, and the death toll passed 160 within a few days.

    Evidence: 19 October 2026 · Drone attacks kill more than 160 in Sudan within days
  • Scenario

    Falsified

    The published claim

    S2 · 30% — Force majeure in Libya: the outage on the Sharara line drags on, the National Oil Corporation declares force majeure and the Zawiya refinery shuts.

    What happened

    The outage did not drag on. The line reopened on 26 September and our records show no force majeure declaration. The loss was contained at 942,376 barrels and about 95 million dollars over 5 days. Only one piece of the scenario materialised: on 26 September the NOC shut one refining unit at Zawiya to protect the second, and the refinery never stopped entirely.

    Evidence: 26 September 2026 · Libya's Sharara–Zawiya pipeline reopens after six days
  • Scenario

    Still open

    The published claim

    S3 · 15% — The Sahel blockade widens: the JNIM blockade reaches beyond the Senegal and Côte d'Ivoire corridors and fuel distribution breaks down in neighbouring countries too.

    What happened

    We found no sourced record of fuel distribution breaking down in neighbouring countries. The Malian army and Africa Corps re-entered Kidal on 6 October, but that says nothing about the direction of the blockade. The data are insufficient for a verdict.

  • Causal chain

    Falsified

    The published claim

    Türkiye pays this spread directly: even if Brent falls to 100.05 dollars, diesel stays above 100 lira in Istanbul and 101 lira in Ankara and Izmir, and the rounds of price rises continue.

    What happened

    Diesel did not stay above 100 lira. A 5.5 lira cut on 24 September brought it down to 90.90 lira on Istanbul's European side, and it rose to 93.45 lira on 25 September; the expected price was 94.55 lira in Ankara and 94.85 lira in Izmir. Price rises kept coming, but the level threshold broke within a week.

    Evidence: 25 September 2026 · Diesel rises by 2.55 lira on 25 September after a 5.5-lira cut on 24 September; a litre climbs to 93.45 lira on Istanbul's European side
  • Watched indicator

    Partly borne out

    The published claim

    A Brent fall below 95 dollars would confirm that physical disruptions are decoupled from price; above 115 dollars is where the supply loss feeds into the global balance.

    What happened

    Brent futures touched neither threshold, at 100.05 dollars on 22 September and 108.79 dollars on 20 October. African disruptions did not pass through to the global price, which fits the thesis. The expectation that prices would ease did not hold, though; the rise was driven by the Gulf, not Africa.

    Evidence: Yahoo Finance (Reuters) — Brent falls as Iran's president discusses mediation

What we learned

We were right that the Libyan outage would be brief and that African disruptions would not pass through to the global price; giving the force majeure scenario 30% was too much. We were wrong in two places. We confined the Sudan drone wave to El Obeid, but the strikes spread to Blue Nile, Darfur and Khartoum. On Türkiye we wrote that diesel would stay above 100 lira, and two days later it fell to 90.90 lira. The same error appears in the Russian refining report published the same day: both reports shared a single assumption about the pump price, and both were wrong together. The lesson: write the Türkiye effect in terms of pass-through speed, not a level threshold.

Published on 22 September 2026 · Reviewed on 21 October 2026

Russia's refining loss has stopped being a war statistic and become a monetary policy line: 45% of capacity is out

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    S1 · 55% — Losses mount and the ban is extended: repairs take weeks, new facilities are hit and the diesel export ban runs beyond 31 October.

    What happened

    Losses mounted. Perm and Novoshakhtinsk were hit on 25 September, Ilsky on 26 September, Volgograd on 2 October, Omsk on 8 October and Ryazan on 15 October. The ban, however, was not extended; it was loosened. On 10 October Russia partly lifted it with a 500,000 tonne tranche under an understanding with the US. The fate of the ban was decided by bargaining with Washington, not by domestic supply, and we did not foresee that.

    Evidence: 10 October 2026 · Russia partly lifts diesel export ban with a 500,000-tonne tranche
  • Scenario

    Falsified

    The published claim

    S2 · 30% — Quick repairs and partial normalisation: damage at the Moscow refinery proves limited, distillation units partly restart within weeks and the ban is lifted at the end of October.

    What happened

    The damage was not limited. According to data we compiled on 19 October, the Moscow refinery is expected to stay offline until early 2027. The IEA measured June runs at 3.8 million barrels a day, a 20-year low, with diesel output down about 30%. The easing of the ban came from a political deal, not from repairs.

    Evidence: 19 October 2026 · The diesel deal's tonnage runs into Russian refining capacity
  • Scenario

    Falsified

    The published claim

    S3 · 15% — An energy infrastructure ceasefire: the sides agree a verifiable arrangement that takes energy infrastructure off the target list, and refinery strikes stop.

    What happened

    No verifiable arrangement emerged. The energy truce Trump declared unilaterally on 11 October was never put in writing. After a 3-day pause Ukraine resumed refinery strikes on 15 October, setting Ryazan alight, and the same night Russia hit 15 regions with more than 850 drones.

    Evidence: 15 October 2026 · Ukraine ends the energy truce as Ryazan refinery comes under fire again
  • Watched indicator

    Falsified

    The published claim

    If Brent keeps easing below 95 dollars, it will strengthen the reading that the loss is concentrated in refining and that crude Russia cannot process is being diverted to exports.

    What happened

    Crude did not ease; it rose. Brent futures, at 100.05 dollars on 22 September, closed at 108.79 dollars on 20 October after touching 111.99 intraday. Crude was pulled higher by attacks on Hormuz and Saudi infrastructure; the assumption that diverting Russian crude to exports would weigh on the price never showed up in prices.

    Evidence: Yahoo Finance (Reuters) — Brent falls as Iran's president discusses mediation
  • Causal chain

    Falsified

    The published claim

    In Türkiye the spread shows up as a product import bill and a falling barrel price does not bring the pump price down; diesel trades above 100 lira.

    What happened

    The pump price came down. A 5.5 lira cut on 24 September took diesel to 90.90 lira on Istanbul's European side, and the 25 September rise brought it to 93.45 lira. Even after a 6.36 lira rise on 10 October it stood at 96.55 lira, still below 100. The barrel price passed through to the pump in both directions.

    Evidence: 10 October 2026 · Turkish diesel jumps over 6 lira as Russia eases its export ban
  • Watched indicator

    Still open

    The published claim

    The Bank of Russia's statement keeps referring to fuel prices and the 14.00% rate is held.

    What happened

    The Bank of Russia's first rate meeting since 11 September falls on 23 October, after the review date. We cannot give a verdict before the decision is published; it will be reviewed again once the statement is out.

What we learned

We were right that refining losses would mount and that there would be no quick repair. We were wrong in three places. We wrote that the ban would be extended because of the domestic supply gap; instead it was loosened early through a diesel deal with the US, meaning our model never included the political channel. The assumption that crude would get cheaper collapsed: Brent futures rose from 100.05 to 108.79 dollars in 4 weeks. The clearest error was on Türkiye. We wrote that diesel would stay above 100 lira; two days later it fell to 90.90 lira and it remains below 100. The lesson: when writing about the product spread, track the lag with which the barrel price reaches the pump rather than a single level threshold.

Published on 22 September 2026 · Reviewed on 21 October 2026

Inside Türkiye's rising confidence index: spending is being pulled forward while the energy shock builds up at the producer

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    S1 · 55% — Front-loading continues and the bill builds up: Hormuz traffic stays low, energy import costs stay high and the disinflation path weakens.

    What happened

    The Hormuz leg held. Kpler recorded only 7 commodity vessels transiting the strait on 6 October, and Gulf crude exports fell to 8.5 million barrels a day in the week before 19 October. The disinflation leg did not. TurkStat said annual inflation fell from 31.51% to 29.73% in September, with the monthly rate of 1.84% coming in below the 2.18% expected.

    Evidence: 5 October 2026 · Softer September inflation puts a rate cut on the table
  • Scenario

    Falsified

    The published claim

    S2 · 30% — The strait normalises: diplomacy advances, Hormuz traffic approaches pre-war levels and the energy import premium recedes.

    What happened

    The strait did not normalise. Transits fell to a two-month low on 6 October, when LSEG showed only 8 crossings. On 19 October Iran expanded its blacklist to 122 ships, up from 46 on 24 August and 77 on 15 September.

    Evidence: 8 October 2026 · Hormuz transits fall to a two-month low
  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: if USDTRY rises above 52, Türkiye's 5-year CDS above 250 basis points and daily Hormuz transits stay below 20, the energy risk premium and a global bloc move are being priced together.

    What happened

    Türkiye's CDS rose from 239.61 on 22 September to 250.08 basis points on 19 October, only just clearing the threshold. The Hormuz threshold held, with PortWatch counting 2–4 transits a day on 8–11 October. The currency threshold is distant: USDTRY was 49.527 on 16 October. Two of the three thresholds were met.

    Evidence: Investing.com — Turkey CDS 5 Years USD
  • Scenario

    Partly borne out

    The published claim

    S3 · 15% — The global bloc move reaches Türkiye: emerging-market risk premia widen in tandem and external financing costs rise without any country-specific news.

    What happened

    Türkiye's premium widened by 10.47 basis points in 4 weeks and some other emerging markets rose too: on 16–19 October Egypt went from 319.54 to 322.57 and Indonesia from 97.26 to 98.01. The 'without country-specific news' condition did not hold, however. Over the same period the fund crisis and the MASAK investigation generated premium through the domestic channel, and on 20 October the heads of the CMB and the POA were removed from office.

    Evidence: 20 October 2026 · Official Gazette removes the heads of Türkiye's market and audit regulators
  • Causal chain

    Still open

    The published claim

    Households bring spending forward in anticipation of price pass-through; in the real sector, export order expectations and investment appetite weaken towards and below the 102.0 level.

    What happened

    October consumer confidence and the Real Sector Confidence Index had not been published by the review date. There is no new data to show which way the September reading of 102.0 is heading; it will be reviewed again when the October figures are released.

What we learned

We were right that Hormuz would stay shut and that the normalisation scenario was weak; the CDS also cleared the 250 threshold on 19 October. We were wrong on the price leg of the main thesis. We wrote that the disinflation path would weaken, yet September inflation came in below expectations and the annual rate fell below 30% for the first time since November 2021. A monthly 0.20% decline in food offset the energy component; we assumed the energy shock would reach the headline too quickly. We attributed the rise in the risk premium to a global bloc move, but much of it came through the domestic channel, from the fund crisis and from decisions targeting the regulators.

Published on 14 October 2026 · Reviewed on 20 October 2026

Black Sea shipping security becomes an annex to the energy bargain

Open the brief ›
  • Falsifier

    Confirmed

    The published claim

    Falsifier: if Trump's 11 October energy truce is put in writing and a pause on ship attacks is folded into the package, our thesis that Black Sea shipping safety is hostage to an energy bargain collapses.

    What happened

    The falsifier never materialised. The truce broke down before anyone drafted it. Zelensky said on 15 October that the ban on strikes against Russian refineries had lapsed; hours later the Ryazan refinery was burning. Russia hit 15 regions the same night with more than 850 drones. We found no Kremlin statement confirming the truce with a date or a scope, and no text containing a pause on ship attacks.

    Evidence: 15 October 2026 · Ukraine ends the energy truce as Ryazan refinery comes under fire again
  • Scenario

    Partly borne out

    The published claim

    S1 · 50% — Unsigned text, continuing strikes: the Kremlin does not confirm the truce, new ship losses occur in the exclusive economic zones, no moratorium is signed, and US CPI prints near 3.6%.

    What happened

    The core held. As of 20 October the moratorium remains unsigned, Ukraine asked Turkey on 16 October to back a corridor for 32 million tonnes of grain exports, and neither party has answered formally. Strikes continued: the Wine Trader burned off Sulina on 15 October with seven crew missing. The location did not hold. Romania places the vessel inside its own territorial sea, so the attack fell not in the economic zone but in the 12-mile belt we had treated as protected. The ship was San Marino-flagged, not Turkish-owned. US CPI came in at 3.7% rather than 3.6%, with core at 2.6% rather than 2.5%.

    Evidence: 15 October 2026 · Seven sailors missing after tanker is hit in Romanian waters
  • Scenario

    Falsified

    The published claim

    S2 · 30% — Shipping annex to the energy deal: Washington puts the refinery-grid truce in writing, a pause on ship attacks is added to the package, and Ankara's text becomes its maritime leg.

    What happened

    No package formed. After a three-day lull Ukraine resumed refinery strikes on 15 October, and Zelensky made Patriot interceptors from Washington a condition for any new pause. The truce was never written down and a shipping annex never reached the agenda. We could not verify any formal response to the maritime security mechanism floated by Turkey's NATO representative.

    Evidence: 15 October 2026 · Ukraine ends the energy truce as Ryazan refinery comes under fire again
  • Scenario

    Still open

    The published claim

    S3 · 20% — EU mission and Montreux friction: the EU adopts Bulgaria's proposal for a Black Sea naval mission at the 15–16 October summit while attacks in the economic zones continue.

    What happened

    The summit's energy session on 15 October ran roughly three hours and ended split over the carbon market. We found no source letting us verify whether a Black Sea naval mission appears in the final communiqué, so we withhold a verdict. This check will be reopened once the text is available.

What we learned

We read the truce correctly: it never reached paper, and the moratorium stayed unsigned. We did not foresee that it would collapse within three days. Our real error was geography. We wrote that attacks were concentrating outside the 12-mile territorial sea and that the coastal corridor was a shelter; two days later a tanker was hit inside Romanian territorial waters and seven crew went missing. Sources put the position at 13 nautical miles, so the boundary carries some ambiguity, but Romania's own determination weakens our assumption. The lesson: treat the insurance exemption as an assumption retested at every incident, not as a structural constant. Our CPI expectation also ran 0.1 point low. The probabilities stand as published.

Published on 14 October 2026 · Reviewed on 20 October 2026

Central banks refuse to rescue, leaving the debt premium with treasuries

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    S1 · 50% — Contained term premium: US CPI prints near 3.6%, the French spread holds at 130–160 basis points, the US 10-year sits in a 5.2–5.4% range and the euro trades between 1.11 and 1.12.

    What happened

    Only one of the three market ranges held. The France-Germany 10-year spread stood near 137 basis points on 16 October, with France at 4.79% and Germany at 3.41%, inside the range. The US 10-year instead fell to 5.088% on 15 October and closed the week at 5.079%, below it. The euro broke above the range at 1.1253 on 16 October. CPI landed close to expectations at 3.7%, yet yields fell even as producer prices jumped to 6.3% year on year, pulled down by safe-haven demand from an AI-driven equity selloff.

    Evidence: 15 October 2026 · US producer inflation jumps to 6.3%, yet Treasury yields fall
  • Watched indicator

    Still open

    The published claim

    Threshold: a US 10-year yield above 5.31% would show the term premium lifting yields on its own as the Fed approaches its peak.

    What happened

    Not breached on a closing basis. After the 14 October CPI print the yield touched 5.33% intraday but closed the day at 5.19%. It then fell to 5.088% on 15 October, the lowest since 23 September, and ended the week down 16.5 basis points at 5.079%. The move went the opposite way to the one we expected. The threshold stays open.

  • Watched indicator

    Still open

    The published claim

    Threshold: a euro below its 12 October low of 1.1155 would show French risk and the ECB-Fed pace gap feeding into the currency together.

    What happened

    Not breached, and the pair moved the other way: 1.1188 on 13 October, 1.1253 on 16 October. The French spread narrowed from roughly 152 to 137 basis points. Moody's reviews France on 23 October; the threshold will be retested after that date.

  • Causal chain

    Still open

    The published claim

    Step 2: as the euro-area term premium rises and the Fed hikes in December, long-dated borrowing costs climb in both euro and dollar, and investors demand a wider premium on emerging-market debt.

    What happened

    The first week ran against the chain: the US 10-year fell 16.5 basis points and the French 10-year about 5. Pricing for a December hike held steady. Warsh gave no signal on 16 October, October odds slid from 33% to roughly 15%, and December sits near 90%. On the emerging side, Turkey's five-year CDS was 245.27 on 14 October and 248.07 on 16 October. The step runs over weeks, so a verdict is premature.

What we learned

We read the French spread correctly at 130–160, but we built the market ranges off the term premium and the Fed, and both misfired. The channel we left out was safe-haven demand: in an AI selloff that took the KOSPI down 7.6% and the Nasdaq down 4.5% on the week, bonds rallied despite a hot PPI print and the euro firmed. The lesson: when writing a yield range, carry equity-driven risk aversion as a separate scenario variable. The thresholds and the chain stay open because they have not matured; they will be retested after Moody's on 23 October and the ECB on 29 October.

Published on 14 October 2026 · Reviewed on 20 October 2026

Saudi supply is back, but the backup outlet sits within missile range

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    S1 · 50% — Fragile recovery holds: Houthi strikes stay confined to administrative and civilian targets, and December Brent trades in a 100–110 dollar range.

    What happened

    The price range held; the target limit did not. Two days later, on 16 October, the Houthis fired a ballistic missile at Yanbu, the export outlet that bypasses Hormuz. The coalition said it downed the missile along with two cruise missiles and three drones. Two tankers were struck off Jubail the same day, so the strikes did not stay confined to administrative targets. Brent closed inside the range at 109.86 dollars on 16 October and 109.55 dollars on 19 October.

    Evidence: 16 October 2026 · The Houthis fire a ballistic missile at Yanbu, Saudi Arabia's Hormuz bypass
  • Scenario

    Still open

    The published claim

    S2 · 30% — Export infrastructure is hit: a pump station on the East-West pipeline, Yanbu or an east-coast terminal takes damage, Saudi exports stall for days and Brent clears 110 dollars.

    What happened

    The trigger came close without firing. Yanbu was targeted and the missile was intercepted; no damage or export disruption was reported. The attacks off Jubail hit vessels rather than the terminal. Brent reached 110.00 dollars intraday on 16 October and 110.70 dollars on 19 October, but closed below 110 on both days.

  • Watched indicator

    Partly borne out

    The published claim

    Threshold: December Brent clearing 110 dollars from its 13 October level of 105.69 would mark the market pricing a shift from administrative buildings to export infrastructure.

    What happened

    The reading was right; the threshold was not cleared on a closing basis. On the day the strike turned to export infrastructure, 16 October at Yanbu, Brent settled 3.72% higher at 109.86 dollars after touching 110.00 intraday. On 19 October it closed at 109.55 dollars with a 110.70 high. The price pressed against the threshold, but an intercepted strike was not enough to carry a close above 110.

    Evidence: Investing.com — Brent Oil Futures Historical Data
  • Watched indicator

    Confirmed

    The published claim

    Threshold: US commercial crude inventories falling below 420 million barrels from 424.1 million on 2 October would show the Gulf shortfall being met out of US stocks.

    What happened

    Cleared on the first print. The EIA reported on 15 October that commercial crude stocks fell 8.6 million barrels in the week to 9 October, to 415.6 million barrels, against a Reuters poll looking for a drop of roughly 0.3 million. The next day Washington opened a 17 million barrel exchange tender from the strategic reserve. How much of the draw came from Gulf-driven export demand cannot be separated out of this release.

    Evidence: Shafaq News — US crude inventories drop 8.6 million barrels

What we learned

The thesis that the backup outlet also sits inside missile range was tested within two days: a ballistic missile was fired at Yanbu. That vindicated the thesis while falsifying the assumption we attached to our most likely scenario, S1, that strikes would stay confined to administrative and civilian targets. Pinning 50% to that assumption was too sanguine. The price range and the 110 dollar threshold held on a closing basis only because the strike was intercepted. The lesson: write an attempt against infrastructure and a hit on infrastructure as two separate triggers. The US inventory threshold was cleared on the first print. The probabilities stand as published.

Published on 13 October 2026 · Reviewed on 18 October 2026

Origin checks stop at the first supplier, while risk builds further down

Open the brief ›
  • Watched indicator

    Falsified

    The published claim

    Secondary indicator: whether the 14 October US CPI and the 15 October CBRT reserve data would show the dollar floor and the energy bill hitting Türkiye in the same week, and whether net reserves excluding swaps would fall below $37.9bn.

    What happened

    There was no pass-through. In the week of 9 October net reserves excluding swaps rose by $6.4bn to $44.2bn, and foreign investors bought $451.1m of government bonds the same week. US September CPI came in at 3.7%, above the 3.6% consensus, yet the dollar index fell from 102.59 on 12 October to 101.84 on 16 October.

    Evidence: 15 October 2026 · CBRT reserves snap a six-week slide with a $4.2bn rise
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: Trump's energy ceasefire is calming energy prices and inflation.

    What happened

    The ceasefire ended on 15 October. Zelenskiy lifted the ban on refinery strikes and fire broke out at the Ryazan refinery. US September CPI rose to 3.7%, with the gasoline index up 9.1% month on month. December Brent futures settled at $109.86 on 16 October.

    Evidence: 15 October 2026 · Ukraine ends the energy truce as Ryazan refinery comes under fire again
  • Watched indicator

    Still open

    The published claim

    Main indicator: an official MOFCOM notice extending the suspension before 10 November, and Chinese customs data on gallium and heavy rare earths by country. A general licence for all buyers, the US included, would disprove the thesis.

    What happened

    The deadline is 10 November. As of 14 October there was no extension notice. On 16 October Beijing added AI models and chip designs to a draft export-control list, so the scope of controls is widening, not narrowing. In September rare earth exports fell 9.8% month on month by tonnage but rose 14.3% year on year by value. The country breakdown has not yet been published.

    Evidence: 16 October 2026 · China moves to bring AI models and chip designs under export controls
  • Scenario

    Still open

    The published claim

    H1 · 55% — Undocumented extension, selective tightening: MOFCOM issues no new notice and the Fed keeps its small, frequent hikes framework; a strong dollar squeezes importers' currencies and Türkiye's external balance.

    What happened

    The horizon has not passed. The absence of a notice fits H1. The dollar leg, however, reversed: the dollar index fell below 102.50 to 101.84 on 16 October, and the ECB reference rate for euro/dollar rose to 1.1253. Warsh gave no rate signal on 16 October and the probability of an October hike fell to 15%.

    Evidence: Investing.com — US Dollar Index Historical Data
  • Scenario

    Still open

    The published claim

    H2 · 25% — Official extension, tension eases: MOFCOM extends the suspension and US September CPI undershoots expectations; the dollar index slips below 102.50.

    What happened

    The horizon has not passed. Neither leg of the trigger arrived: there is no official extension, and CPI beat expectations at 3.7%. Even so, H2's market outcome materialised, with the dollar index easing to 101.84 on 16 October. Inflation did not drive the fall; a flight into bonds after the AI-led equity sell-off on 15 October did.

    Evidence: 15 October 2026 · US producer inflation jumps to 6.3%, yet Treasury yields fall
  • Falsifier

    Still open

    The published claim

    Falsifier (30 November): the reading that risk premia have been written into contracts is disproved if the London insurance market narrows the Black Sea high-risk area to its pre-September extent, VLCCs turned back from Fujairah return within two weeks, and Yandex withdraws its liability exclusion clause.

    What happened

    The deadline is 30 November. There is no sign of narrowing. The Turkish-owned Azov Concorde sank off Bulgaria after a drone strike on 13 October, and a tanker was hit in Romanian territorial waters on 15 October. We found no open data showing VLCCs returning to Fujairah.

    Evidence: 13 October 2026 · Turkish-owned Azov Concorde sunk by drones off Bulgaria

What we learned

This is the second time we made the same mistake on the Türkiye line, after the 12 October brief. We implied that the dollar floor and the energy bill would push reserves down in the same week. Instead net reserves excluding swaps rose by $6.4bn, and the dollar index fell to 101.84 despite strong CPI. We assumed hot inflation would carry the dollar and ignored the safe-haven bid for bonds that the AI sell-off created. The core thesis, the lower-layer constraint and the 10 November suspension, has not yet reached its deadline. From now on we will not imply a direction on the reserve line without weighing portfolio and tourism inflows.

Published on 13 October 2026 · Reviewed on 18 October 2026

As the Fed tightens, Washington steps away from the debt table

Open the report ›
  • Causal chain

    Partly borne out

    The published claim

    Step 1: if the 14 October CPI meets or beats the 3.6% consensus, the market pulls the Fed's next hike forward to the 27–28 October meeting and the 2-year yield rises above 4.80%.

    What happened

    The condition was met, but the outcome did not hold. CPI came in at 3.7% and the FedWatch probability of an October hike jumped from 22% to 39%. The 2-year yield closed 14 October at 4.802% after touching 4.952% intraday. Two days later the picture reversed: Warsh gave no signal, the October probability fell to 15% and the 2-year yield eased to 4.736% on 16 October.

    Evidence: Investing.com — U.S. 2 Year Treasury Historical Data
  • Causal chain

    Falsified

    The published claim

    Step 2: as long as the rate differential stays wide, the dollar strengthens; the euro and emerging-market currencies stay under pressure and central banks spend reserves.

    What happened

    The dollar weakened. The dollar index fell from 102.59 on 12 October to 101.84 on 16 October. On the ECB reference rate, euro/dollar rose from 1.1181 to 1.1253 over the same period, moving away from the report's 1.1155 threshold. The CBRT's net reserves excluding swaps rose by $6.4bn in the week of 9 October.

    Evidence: Investing.com — US Dollar Index Historical Data
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: the energy ceasefire will ease inflationary pressure.

    What happened

    The pressure did not ease. September CPI rose to 3.7%, with the energy index up 5.2% and gasoline up 9.1% month on month. The next day PPI rose 1.3% on the month, its biggest gain in four and a half years. The ceasefire ended on 15 October.

    Evidence: 14 October 2026 · US inflation climbs to 3.7%, putting an October hike back in play
  • Scenario

    Still open

    The published claim

    H1 · 55% — A Fed moving in frequent steps and a strong dollar: the dollar index above 102.50, euro/dollar in a 1.11–1.12 band and the US 10-year yield between 5.2% and 5.4%.

    What happened

    The horizon is 0–3 months and has not passed. Yet all three market legs currently sit outside their bands: the dollar index was 101.84 on 16 October, euro/dollar 1.1253, and the 10-year yield ended the week at 5.079%. Half the trigger arrived (CPI at 3.7%) and half did not, as Warsh lent no support to Hammack's framework.

    Evidence: 16 October 2026 · Warsh gives no rate signal in Bangkok and October hike odds fall
  • Scenario

    Still open

    The published claim

    H3 · 20% — The debt premium takes the lead: the US 10-year yield breaks above 5.31% while the dollar weakens; gold rises above $4,200.

    What happened

    The horizon has not passed. The 10-year yield hit 5.33% intraday on 14 October but closed at 5.19%, then fell to 5.08% on 15 October. The dollar weakened, but yields fell too, so H3's divergence did not form. Gold stood at $4,170.39 on 15 October, below the $4,200 threshold.

    Evidence: 15 October 2026 · US producer inflation jumps to 6.3%, yet Treasury yields fall

What we learned

We built the chain on a single channel: hot CPI, a hawkish Fed, a strong dollar. CPI beat expectations and PPI hit a four-and-a-half-year high, yet the dollar retreated. Inflation was not what broke the chain. On 15 October the KOSPI fell 7.6%, the AI-led sell-off created safe-haven demand for bonds and the 10-year yield dropped to 5.08%. We were also wrong on the Fed: Warsh gave no signal and the market pushed the hike from October to December. We had defined the chain-breaking condition only in terms of data and Fed speeches, and never wrote in a risk-appetite shock. In dollar chains we will now list safe-haven bond demand as a separate breaker.

Published on 12 October 2026 · Reviewed on 18 October 2026

Empty tankers are turning away from Fujairah, the Gulf's last buffer

Open the report ›
  • Causal chain

    Falsified

    The published claim

    Step 1: shuttle tankers that cannot find larger vessels are unable to discharge, so Gulf crude exports fall even if the strait stays open (about 10 million barrels a day on 7 October, according to Kpler).

    What happened

    Exports did not fall. TankerTrackers data cited by Reuters put Middle East crude exports at an average of 17.273 million barrels a day in October, roughly 2.7 times the May level; Kpler puts the pre-war average at 18 million. According to the UAE, Hormuz flows rose by 3.7 million barrels a day in September to 10.9 million, with the increase carried by transfers in the Gulf of Oman.

    Evidence: 13 October 2026 · Gulf crude exports top September despite mounting Hormuz attacks
  • Watched indicator

    Falsified

    The published claim

    Flagged as noise: Saudi supply continues come what may, and Gulf exports have recovered.

    What happened

    The data confirmed the recovery. Saudi Arabia reported September crude output of 8.37 million barrels to OPEC. According to the UAE, Gulf crude exports exceeded 75% of pre-war levels in September, and October flows are running above September. The risk persists: the Houthis fired a ballistic missile at the Yanbu outlet on 16 October. But treating the volumes as noise was wrong.

    Evidence: 13 October 2026 · Saudi crude output recovers to 8.37 million barrels in September
  • Causal chain

    Partly borne out

    The published claim

    Chain-breaking condition: if the number of empty VLCCs heading to Fujairah rises again and Kpler's Gulf crude exports exceed 12 million barrels a day, the transfer hub is holding.

    What happened

    The export leg was met: TankerTrackers' October average of 17.273 million barrels a day is well above the threshold. We could not access Kpler's weekly series from open sources. Nor did we find data showing VLCCs returning to Fujairah, so that leg could not be verified.

    Evidence: 13 October 2026 · Gulf crude exports top September despite mounting Hormuz attacks
  • Scenario

    Still open

    The published claim

    H2 · 30% — The transfer hub breaks: new hits off Fujairah and Oman or an attack on the East–West line; Gulf exports drop below 10 million barrels, December Brent futures pass $110 and VLCC rates hit a record.

    What happened

    The horizon is 0–3 months. The trigger arrived in part: a vessel was hit at Oman's Sohar anchorage on 12 October, and a missile was fired at the Yanbu outlet on 16 October. December Brent futures touched $110.00 intraday on 16 October and settled at $109.86. Rates did not set a record, as the Baltic index fell from 841 to 776 Worldscale. Exports did not fall below 10 million.

    Evidence: 16 October 2026 · The Houthis fire a ballistic missile at Yanbu, Saudi Arabia's Hormuz bypass
  • Scenario

    Still open

    The published claim

    H1 · 50% — The buffer erodes and flows stay tight: Gulf crude exports hold in a 10–12 million barrels a day band and ICE December Brent futures stay within $100–110.

    What happened

    The horizon is 0–3 months. The price leg is holding: December futures settled between $105.69 and $109.86 from 12 to 16 October. The flow leg is not: October exports are well above the 10–12 million band. According to the UAE, onshore inventories fell by 113 million barrels in September; the buffer is eroding in stocks, not in flows.

    Evidence: Investing.com — Brent Oil Futures Historical Data (LCOZ6)

What we learned

We built the report's premise on a single day's reading: Kpler's figure of about 10 million barrels on 7 October. The October average came in at 17.3 million barrels a day, and empty tankers turning back from Fujairah did not cut exports. We treated the Saudi recovery as noise; the data showed it was real. The price thesis still stands: December futures touched $110 on 16 October and onshore inventories are at their lowest since 2016. The constraint lies in stocks, not volumes. From now on we will not present a single day as a trend in a premise; a weekly average at minimum and a second tracking source are required.

Published on 12 October 2026 · Reviewed on 17 October 2026

Oil's bottleneck has moved from the strait to the tanker fleet

Open the brief ›
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: the energy ceasefire Trump announced on 11 October halted the Russia–Ukraine energy war.

    What happened

    The ceasefire lasted four days. On 15 October Zelenskiy announced that the ban on strikes against refineries had been lifted, and hours later a fire broke out at the Ryazan refinery. According to Al Jazeera, Russia launched more than 850 drones on the night of 15 October and struck 15 regions. Russia had kept attacking the grid during the three-day interval as well.

    Evidence: 15 October 2026 · Ukraine ends the energy truce as Ryazan refinery comes under fire again
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: Middle East crude exports have surpassed pre-war levels and the supply problem is solved.

    What happened

    According to the International Energy Agency's (IEA) report of 14 October, Gulf crude exports rose to 13.9 million barrels a day in September, but reached only 75% of the pre-war level. The region's product exports stood at 1.9 million barrels a day, against 4.7 million before the war. Global observed stocks are 509 million barrels below pre-war levels; the agency expects the diesel shortage to extend into the second half of 2027.

    Evidence: 14 October 2026 · IEA sees Gulf flows recovering as global stock deficit hits 509m barrels
  • Watched indicator

    Partly borne out

    The published claim

    Key indicator: if UKMTO's weekly Hormuz count stays below 300 transits and the Gulf–East Asia VLCC rate holds above $1 million a day, the ship shortage is becoming entrenched.

    What happened

    The transit leg held. According to JMIC, there were 121 US-escorted transits on 11–14 October, roughly 30 a day; that pace stays below 300 a week. We could not verify the rate leg: we found no daily dollar rate in open sources. Baltic Exchange's Gulf of Oman–China index, meanwhile, fell from 841 to 776 Worldscale this week. The rate has come off its record but remains close to it.

    Evidence: 14 October 2026 · Container ship GFS Pride sinks after being struck in Hormuz
  • Scenario

    Still open

    The published claim

    H1 · 50% — The ship shortage becomes entrenched: the spread between spot Brent and the December contract stays above $20 and tanker freight remains elevated.

    What happened

    The three-month horizon has not expired. The spread is holding at the threshold: on FRED, spot Brent was $126.55 on 14 October and the December contract $106.06 the same day, a gap of $20.49. That is about $4.4 narrower than the $24.86 of 6 October. On 14 October Washington began debating an easing of sanctions on shadow fleet tankers; this is a policy lever aimed directly at freight, and one our scenarios left out.

    Evidence: 14 October 2026 · US weighs easing sanctions on shadow fleet tankers
  • Watched indicator

    Falsified

    The published claim

    Türkiye constraint: a freight-driven energy bill narrows the CBRT's room to cut through the reserve channel; the 15 October data would show whether net reserves excluding swaps fell below $37.9 billion.

    What happened

    The first data point showed the opposite. In the week of 9 October, net reserves excluding swaps rose by $6.4 billion to $44.2 billion, and gross reserves reached $171.575 billion. The August current account, released on 13 October, posted a $4.053 billion surplus on the back of tourism. In the Matriks survey, 32 of 37 economists expect a 100-basis-point cut on 22 October; for now the reserve channel is not narrowing the room to cut.

    Evidence: 15 October 2026 · CBRT reserves snap a six-week slide with a $4.2bn rise
  • Falsifier

    Still open

    The published claim

    Falsifier (31 October): if Ukraine keeps striking Russian refineries and the US does not suspend PURL deliveries, or if Russia fails to ship the 500,000-tonne November tranche, the reading that treats the energy ceasefire as a precondition for diesel supply fails.

    What happened

    The deadline is 31 October. The first half of the falsifier has started to play out: Ukraine resumed refinery strikes on 15 October. There is no report of PURL being suspended; if no suspension comes, our reading weakens. The view that Washington's pressure was tied to the diesel deal found support: the OFAC licence was issued on 9 October, and 12 senators called for its withdrawal on 13 October.

    Evidence: 13 October 2026 · Twelve US senators urge Trump to scrap the Russian diesel deal

What we learned

Both our noise labels held: the ceasefire collapsed within four days, and Gulf volumes did not return to pre-war levels. Our error lies in the Türkiye line. We implied that the freight bill would push reserves down and narrow the room to cut. In the first week, net reserves excluding swaps rose by $6.4 billion and the August current account posted a $4 billion surplus. We drew a conclusion from a single cost channel without accounting for tourism revenue and portfolio inflows. A second gap: our scenarios never considered that the US could target freight directly by easing sanctions. We cannot reach a daily dollar series for ship rates in open sources; from now on we will also define the threshold using Baltic Exchange's Worldscale index.

Published on 10 October 2026 · Reviewed on 17 October 2026

Real rates, not inflation, drive US yields higher and lift the dollar

Open the report ›
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: US bond yields are rising on inflation fears; 10-year breakeven inflation is flat at 2.35%, and the whole increase comes from the real yield.

    What happened

    It passed the hardest test. September CPI came in at 3.7% year on year on 14 October, and PPI at 6.3% on 15 October; the breakeven rate was 2.35% on 14 October, 2.32% on 15 October and 2.35% on 16 October. The 10-year yield fell from 5.24% to 5.10% between 9 and 15 October, and the real yield from 2.91% to 2.78%. Once again, almost all of the move came from the real yield.

    Evidence: 15 October 2026 · US producer inflation jumps to 6.3%, yet Treasury yields fall
  • Scenario

    Still open

    The published claim

    H1 · 55% — The real-rate regime persists: the 10-year real yield stays in a 2.8–3.0% range, the US 10-year yield in a 5.2–5.4% band and the broad dollar index between 121 and 123.

    What happened

    The three-month horizon has not expired, but two of the bands broke within a week. According to FRED, the 10-year yield dropped below its band at 5.10% on 15 October, and the real yield below its range at 2.78%. The broad dollar index was inside its range at 121.60 at the latest observation, on 9 October. The fall came not from the Fed but from safe-haven demand following the KOSPI's 7.6% crash.

    Evidence: FRED — 10-Year Treasury Inflation-Indexed Security, Constant Maturity (DFII10)
  • Causal chain

    Falsified

    The published claim

    Step 1 (weeks): the real-rate gap drives flows into the dollar and emerging-market central banks spend reserves; watch whether the CBRT's net reserves excluding swaps fall below $35 billion.

    What happened

    The first weekly data came in the opposite direction. According to CBRT data released on 15 October, net reserves excluding swaps rose by $6.4 billion to $44.2 billion in the week of 9 October, and gross reserves rose by $4.187 billion to $171.575 billion. Far from nearing the $35 billion threshold, the buffer grew.

    Evidence: 15 October 2026 · CBRT reserves snap a six-week slide with a $4.2bn rise
  • Causal chain

    Still open

    The published claim

    Step 2 (weeks): because lira carry demand competes with the US real yield, the CBRT's room to cut narrows and foreign selling of local government bonds continues.

    What happened

    The MPC meets on 22 October; the room to cut has not yet been tested. The testable half did not hold: non-residents, who sold $484.7 million of government bonds the previous week, bought $451.1 million in the week of 9 October. Equity selling of $248.2 million continued. In the Matriks survey, 32 of 37 economists expect a 100-basis-point cut.

    Evidence: 15 October 2026 · CBRT reserves snap a six-week slide with a $4.2bn rise
  • Causal chain

    Still open

    The published claim

    Chain-breaking condition: if the 10-year real yield returns below its mid-September level of 2.62% and the broad dollar index drops below 119, the chain breaks at the first step.

    What happened

    The condition has not been met. The real yield was 2.78% on 15 October, 16 basis points above the threshold; the broad dollar index was 121.60 on 9 October, well above 119. The direction, however, points towards the threshold: the real yield has eased 17 basis points from its 5 October peak of 2.95%.

    Evidence: FRED — Nominal Broad U.S. Dollar Index (DTWEXBGS)

What we learned

The diagnosis held: even as inflation data showed the sharpest PPI rise in four years, the breakeven rate stayed at 2.32–2.35% and yields moved with the real component. Where we went wrong was the transmission chain. We wrote that a strong dollar would make the CBRT spend reserves and push foreigners out of local government bonds. In the first week, net reserves excluding swaps rose by $6.4 billion and foreigners put $451.1 million into those bonds. We tied the global rate channel to Türkiye mechanically, without weighing domestic carry returns or the volatility of weekly flows. One week of data does not conclusively refute a step, but the $35 billion threshold is now $9 billion away.

Published on 12 October 2026 · Reviewed on 17 October 2026

Oil's binding constraint is now ships, not barrels

Open the report ›
  • Scenario

    Still open

    The published claim

    H2 · 30% — The transfer zone becomes a target: hits continue in the Gulf of Oman transfer area and off Fujairah; the VLCC rate sets a new record and the ICE Brent December contract tests $110.

    What happened

    The trigger arrived, but the market did not move the way we expected. According to UKMTO, six ships, five of them tankers, were hit in a single day on 12 October; one was at Oman's Sohar anchorage. GFS Pride sank on 14 October. Rates, however, set no record: Baltic Exchange's Gulf of Oman–China index fell from 841 to 776 Worldscale this week. The Brent December contract stayed in a $105.69–106.65 range on 12–15 October.

    Evidence: 13 October 2026 · Gulf crude exports top September despite mounting Hormuz attacks
  • Watched indicator

    Confirmed

    The published claim

    Threshold: while the gap between EIA spot Brent and the ICE Brent December contract stays above $10, the physical crude premium is carrying freight; on 6 October the gap was $24.86.

    What happened

    The gap stayed at more than twice the threshold. According to FRED, spot Brent closed at $126.55 on 14 October and the December contract at $106.06 the same day, a gap of $20.49. That is $4.37 narrower than on 6 October.

    Evidence: FRED — Crude Oil Prices: Brent - Europe (DCOILBRENTEU)
  • Watched indicator

    Partly borne out

    The published claim

    Allocation indicator: the path of the Gulf–East Asia VLCC daily rate relative to the $1 million level; it set a record of $1.4 million a day on 7 October.

    What happened

    The rate retreated from its record; its level could not be verified. According to a Reuters report of 16 October, the Baltic Exchange index fell from 841 a week earlier to about 776 Worldscale. Analysts attributed the drop to demand weakening after fresh attacks and to reports that shadow fleet sanctions might be eased. No daily dollar equivalent is available in open sources.

    Evidence: gCaptain (Reuters) — Supertanker Rates Fall from Record Highs
  • Causal chain

    Still open

    The published claim

    Chain-breaking condition: if UKMTO's weekly Hormuz count rises from 275 transits to above 500, the need for transfers falls, the VLCC rate drops and the chain stops at the first step.

    What happened

    The condition has not been met. According to JMIC, there were 121 US-escorted transits on 11–14 October, roughly 30 a day; the pre-war average was 138 a day. CENTCOM counted more than 40 ships in one 24-hour period, and even that pace does not reach 300 a week. Rates fell without transits rising; what broke the chain was not transits but demand and the sanctions news.

    Evidence: 14 October 2026 · Container ship GFS Pride sinks after being struck in Hormuz

What we learned

In H2 we built one mechanism the wrong way round. We wrote that attacks on the transfer zone would deepen the ship shortage and push rates to a record. According to Reuters, tanker demand weakened after the hits of 12–14 October and rates fell from 841 to 776 Worldscale. Attacks curb supply, but they also curb demand for ships; the report never weighed this second effect. We also left out of our scenarios a policy lever aimed directly at freight: easing sanctions on the shadow fleet. The physical–futures gap thesis still stands, at $20.49.

Published on 6 October 2026 · Reviewed on 16 October 2026

Food pulled headline inflation down; reserves hold the key to rates

Open the report ›
  • Watched indicator

    Falsified

    The published claim

    H2 trigger: the 8 and 15 October reserve releases show the drain in gross reserves extending into a seventh and eighth week.

    What happened

    The trigger did not fire. The official release on 8 October covered not a new week but the week of 2 October, which the report had already counted; gross reserves fell to $167.388 billion. The release on 15 October, covering the week to 9 October, broke the run: gross reserves rose by $4.187 billion to $171.575 billion, and gross FX reserves rose by $4.241 billion to $64.567 billion. Gold reserves slipped by $54 million.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Watched indicator

    Falsified

    The published claim

    Flagged as signal: the risk premium is climbing alongside reserves; Türkiye's 5-year CDS stood at 220.50 basis points on 21 August and 255.69 on 2 October.

    What happened

    Reserves kept falling in the week of 2 October, but CDS did not move with them. CDS eased from 255.29 on 5 October to 244.07 on 9 October and 243.28 on 12 October, then stood at 245.27 basis points on 14 October. The simultaneous link we drew between reserves and the risk premium did not hold in this window.

    Evidence: Investing.com — Turkey CDS 5 Years USD historical data
  • Scenario

    Still open

    The published claim

    H1 · 45% — Limited cut, tight liquidity: the CBRT cuts by 100 basis points on 22 October; USD/TRY stays in a 49–50 band and Türkiye CDS fluctuates between 250 and 270 basis points.

    What happened

    The MPC meets on 22 October, so the scenario has not matured. The interim picture is mixed. USD/TRY was inside the band at 49.5278 on 15 October. Türkiye CDS, however, traded at 243.28–245.27 basis points between 9 and 14 October, below the 250–270 range the scenario envisaged.

    Evidence: Investing.com — USD/TRY historical data
  • Falsifier

    Still open

    The published claim

    Chain-breaking condition: if Brent falls durably below $90, or if the 8 and 15 October reserve releases show the drain stopping and portfolio inflows starting, the chain breaks at the first step.

    What happened

    Half the condition has been met: the reserve drain stopped in the week to 9 October, with gross reserves up $4.187 billion. We could not confirm from the securities data for the week to 9 October that portfolio inflows had started. Brent closed at $105.92 on 15 October, well above the $90 threshold. The first step of the chain hinges on the 22 October decision.

    Evidence: Euronews — CBRT reserves rise to $171.6 billion

What we learned

We were wrong in two places. We assumed the 8 October release covered a new week, when it was the week of 2 October that the report had already used. Misreading the data calendar led us to set the H2 trigger a week too far ahead. We also treated reserves and CDS as a single signal rising in tandem. Between 5 and 14 October CDS fell by more than 10 basis points, while reserves fell for one more week and then turned. Change of method: for weekly data we will first state the week covered, and before tying two indicators to one signal we will check the latest reading of both.

Published on 6 October 2026 · Reviewed on 16 October 2026

Oil's squeeze moves from crude to products, and the Mediterranean pays

Open the report ›
  • Watched indicator

    Falsified

    The published claim

    Flagged as noise: Hormuz attacks are pushing oil above $100; Brent closed at $100.32 on 5 October and eased to $98.69 in the 6 October table.

    What happened

    Brent did not stay below $100. The Investing.com series shows $100.20 on 7 October and $104.28 on 8 October. After two outbound tankers were hit on 10 and 11 October, it reached $106.65 on 12 October. The price held in a $105–107 range through 15 October. Reuters wrote on 15 October that the conflict in the Middle East was keeping the geopolitical risk premium alive.

    Evidence: 11 October 2026 · Second outbound tanker hit in Hormuz in two days
  • Causal chain

    Partly borne out

    The published claim

    Step 2 (days): with the sliding-scale tax mechanism ending on 1 October, imported product costs pass through to the pump; diesel stays high even if Brent eases.

    What happened

    Pass-through came as quickly as we expected. On 10 October diesel rose by 6.36 lira a litre, taking the price on Istanbul's European side to 96.55 lira. The second half of the claim could not be tested: Brent did not ease over the period, rising from $100.20 on 7 October to $106.65 on 12 October. It remains unclear whether the increase came from the product premium or from crude.

    Evidence: 10 October 2026 · Turkish diesel jumps over 6 lira as Russia eases its export ban
  • Scenario

    Still open

    The published claim

    H1 · 55% — Crude ample, products scarce: Gulf crude exports stay above 90%, the Russian diesel ban extends into November, and Brent trades sideways around $100.

    What happened

    The scenario has not matured, but two components are diverging. On 10 October Russia partly lifted its diesel ban with a 500,000-tonne tranche, so an unchanged ban running into November now looks unlikely. Brent did not trade sideways around $100; it closed between $105.69 and $106.65 on 12–15 October. Gulf crude flows, meanwhile, exceeded September's level in October.

    Evidence: 10 October 2026 · Russia partly lifts diesel export ban with a 500,000-tonne tranche
  • Watched indicator

    Partly borne out

    The published claim

    Constraint: the safe route around Hormuz runs through a single pipeline; Kpler puts East-West flows at about 2.65 million barrels a day, against 5.5 million before the attack.

    What happened

    The single-line constraint persists, but our flow figure clashed with another measurement published the same day. According to a report based on Bloomberg, the line was carrying 5.8 million barrels a day on 6 October after repairs, and Riyadh is raising capacity from 7 to 10 million barrels. Kpler's port loadings and pipeline throughput measure different things; the report did not point out that difference.

    Evidence: 12 October 2026 · Saudi Arabia plans 10 million barrels of pipeline capacity bypassing Hormuz
  • Watched indicator

    Still open

    The published claim

    Threshold: an ICE Brent front-month close above $110 would signal a move to H2, a second shock with no buffer.

    What happened

    The threshold was not crossed. The highest close was $106.65 on 12 October; Brent settled 14 cents lower at $105.92 on 15 October. The Houthi strike on Aramco's building in Dhahran on 13 October does not count as a pipeline or Yanbu outage.

    Evidence: Yahoo Finance (Reuters) — Brent oil slips as US weighs easing sanctions on global shadow fleet

What we learned

Our clearest error was the noise label. We wrote that Hormuz attacks were not pushing the price above $100; two days later Brent hit $104, six days later $106, and it stayed there. We took a single day's fall as proof and dismissed the channel as noise too early. For the East-West line we presented one data provider's port figure as pipeline flow, and did not mention the conflicting measurement even though it was published the same day. Change of method: the noise label will require at least a week of price behaviour, and every flow figure will state what kind of measurement it is.

Published on 9 October 2026 · Reviewed on 14 October 2026

Investors keep credit risk but shun duration and currency

Open the brief ›
  • Falsifier

    Still open

    The published claim

    Falsifier (31 October): no new tanker or terminal attack occurs around Fujairah and Yanbu, and bypass exports stay above 6.7 million barrels a day. If IDCF also restores data in its four regions from its own systems, the reading that the backup route sits in the same threat zone as the main route is refuted.

    What happened

    Due 31 October. The first leg of the falsifier has not held: Argus data show two more tankers hit on 11 October, one of them off Oman. Bloomberg's count on 12 October shows at least 10 empty supertankers bound for Fujairah stopped or turned back. A drop in bypass flows below 6.7 million barrels has not yet been measured.

    Evidence: 12 October 2026 · Empty supertankers turn back from the Fujairah transshipment hub
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: Trump's 8 October message of no strike before 3 November did not reduce oil risk.

    What happened

    After the message, the December Brent contract rose from $104.28 on 8 October to $106.65 on 12 October and settled at $105.69 on 13 October. Two outbound tankers were hit in Hormuz on 10–11 October.

    Evidence: Investing.com — Brent oil futures historical data
  • Scenario

    Still open

    The published claim

    H1 · 50% — Carry demand weak but financed: in the 15 October data, outflows from government bonds stay below $500 million, net reserves excluding swaps hold in a $37–40 billion band, and Türkiye CDS stays in a 235–260 band.

    What happened

    Due 15 and 22 October. The market leg is within the band for now: Türkiye's 5-year CDS stood at 243.28 basis points on 12 October. Reserve and securities data for the week of 9 October have not yet been published.

    Evidence: Investing.com — Turkey CDS 5-year USD historical data
  • Scenario

    Still open

    The published claim

    H3 · 20% — External rates ease and carry returns: US CPI on 14 October comes in below expectations, the US 10-year falls below 5.10% and December Brent drops back below $100.

    What happened

    The trigger has not come. December Brent was above $100 at $105.69 on 13 October. US September CPI is due at 15:30 on 14 October.

    Evidence: Investing.com — Brent oil futures historical data

What we learned

Most verdicts are not yet due. On oil, the noise call held: the peace message did not lower the price. Measurement correction: the previous night we recorded the 12 October Brent close as $106.12. The Investing.com table now shows $106.65 for that day, matching the ICE settlement price. $106.12 was an intraday value; the sourced data have been corrected. We will take closes from the table only the following day.

Published on 9 October 2026 · Reviewed on 14 October 2026

Europe exempted defence from its debt rules, and the term premium pays

Open the report ›
  • Watched indicator

    Still open

    The published claim

    Threshold: the German 30-year yield rising from 3.84% on 8 October to above 4% would show defence debt carrying the term premium into the core.

    What happened

    The yield moved away from the threshold. In the Investing.com series the German 30-year was 3.859% on 9 October, 3.839% on 12 October and 3.810% on 13 October.

    Evidence: Investing.com — Germany 30-Year Bond Yield historical data
  • Scenario

    Still open

    The published claim

    H1 · 50% — The rule bends and the premium becomes entrenched: the German 30-year yield stays above 3.7% and the France–Germany spread in a 130–160 basis-point band.

    What happened

    The German part of the market leg holds: 3.810% on 13 October, above 3.7%. The French budget vote and the Commission decision have not yet come; the spread was not measured in this check.

    Evidence: Investing.com — Germany 30-Year Bond Yield historical data
  • Watched indicator

    Still open

    The published claim

    Threshold: EUR/USD falling from around 1.121 on 8 October to below 1.10 would show fiscal strain passing into the currency channel.

    What happened

    Not triggered. The ECB reference rate was 1.1181 on 12 October.

What we learned

Not yet due. The report pointed to upward pressure at the far end of the curve, but in the first three trading days the German 30-year yield fell 5 basis points. That is not yet an error; H1's 3.7% floor holds. Still, the 4% threshold looks distant in the near term. We will measure the France–Germany spread directly from a source in the next check.

Published on 9 October 2026 · Reviewed on 14 October 2026

Iranian oil hinges on reservoirs and tankers, not on a peace deal

Open the report ›
  • Watched indicator

    Still open

    The published claim

    Threshold: if December Brent falls below $95, the market is pricing a rapid return of supply including Iran, in other words a peace that ignores the constraint.

    What happened

    Not triggered. December Brent was $106.65 on 12 October and $105.69 on 13 October, about $11 above the threshold.

    Evidence: Investing.com — Brent oil futures historical data
  • Scenario

    Still open

    The published claim

    H1 · 55% — Extended blockade: no deal before 3 November, the blockade and sanctions continue, and Iranian loadings stay at zero.

    What happened

    Due 3 November; interim direction is towards H1. On 11 October Iran found the US response to its seven conditions insufficient and said it had cut off correspondence. On 10 October the US hit a cargo ship trying to run the blockade; CENTCOM says 135 ships have been turned back.

    Evidence: 11 October 2026 · Iran rejects US reply on Hormuz terms and halts exchanges
  • Scenario

    Still open

    The published claim

    H3 · 15% — Escalation: attacks on Fujairah or Red Sea terminals cut the bypass flow of 6.7 million barrels a day.

    What happened

    Close to the trigger but not reached. No terminal hit has been confirmed. Bloomberg data show at least 10 empty supertankers bound for Fujairah stopped or turned back by 12 October; the transshipment system is slowing. Kpler put Gulf crude exports at 10 million barrels a day on 7 October.

    Evidence: 12 October 2026 · Empty supertankers turn back from the Fujairah transshipment hub
  • Watched indicator

    Still open

    The published claim

    Threshold: if daily Hormuz transits exceed 20 for a week, the strait is effectively reopening.

    What happened

    Not triggered. In the IMF PortWatch series, daily transits were in a 0–4 range between 28 September and 4 October. UKMTO says traffic in the seven days to 9 October was about 75% below pre-war levels.

    Evidence: 10 October 2026 · US warplane strikes cargo ship trying to run Iran blockade

What we learned

No threshold has been tested and nothing is yet due; the interim course is consistent with the report's main scenario. There is one risk we flag separately: we tied H3 solely to a terminal attack. Yet the first mechanism slowing the bypass was not a hit but empty tankers not going to Fujairah. We should have tied the trigger to flow volume rather than physical damage.

Published on 9 October 2026 · Reviewed on 14 October 2026

Foreigners are quitting the lira, not Türkiye, and carry limits the cut

Open the report ›
  • Watched indicator

    Still open

    The published claim

    Threshold: if the 10-year yield rises from 32.84% on 8 October above 34%, the market has started pricing lira risk rather than a cut.

    What happened

    Not crossed. In the Investing.com series the 10-year yield was 32.75% on 12 October and 32.80% on 13 October. Trading Economics also gives 32.80% for 13 October. The 35.495% seen the previous night no longer appears in the table.

    Evidence: Investing.com — Turkey 10-Year Bond Yield Historical Data
  • Watched indicator

    Still open

    The published claim

    Threshold: if CDS rises from 247.80 on 7 October above 275 basis points, lira concern has jumped into credit risk.

    What happened

    Not crossed; it fell instead. Türkiye's 5-year CDS was 244.07 basis points on 9 October and 243.28 on 12 October.

    Evidence: Investing.com — Turkey CDS 5-year USD historical data
  • Watched indicator

    Still open

    The published claim

    Threshold: if USD/TRY rises from its 49.22–49.35 band on 8 October above 50.50, reserve defence can no longer hold the currency.

    What happened

    Not crossed. In the series derived from ECB reference rates, USD/TRY was 49.337 on 12 October, at the top of the band.

  • Scenario

    Still open

    The published claim

    H1 · 45% — A cut without carry: the MPC cuts by 100 basis points on 22 October, foreigners keep exiting government bonds gradually, and net reserves excluding swaps fall into a $35–37 billion band.

    What happened

    Due 22 October. Reserve and government-bond data for the week of 9 October will be published at 14:30 on 15 October; the verdict will follow those data and the MPC decision.

What we learned

Not yet due. All three market thresholds are far off; the lira risk in the report has not yet shown up in the currency or CDS prices. Where we were wrong is measurement: the previous night we wrote that the Investing.com 10-year series had switched to compound yields on 12 October, and we suspended the indicator. Today the same table shows 12 October at 32.75%, and a second source agrees. Either we saw a temporary data error or the series was revised back. From now on we will judge a series change not on a single observation but with two sources over two consecutive days.

Published on 8 October 2026 · Reviewed on 13 October 2026

The floor under borrowing costs rises as Türkiye borrows to pay interest

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    Primary watch: if the indirect share at the 22 billion dollar US 30-year auction on 8–9 October stays close to the 80.3% seen at the 10-year and the dealer share stays below 10%, the floor is high but financed; if it slides towards 57.2%, dealers end up carrying inventory.

    What happened

    The 8 October auction stopped at a high yield of 5.618%, with a bid-to-cover ratio of 2.54. The indirect share of 72.3% landed between the two thresholds: 8 points away from 80.3%, but nowhere near 57.2%. The dealer share of 6.8% was below 10%, yet roughly 3 times the 2.5% at the 10-year. The floor was financed, but at a cost 31 basis points higher than the 5.308% at the September auction.

    Evidence: 8 October 2026 · Weak demand at US 30-year auction breaks with strong 10-year sale
  • Watched indicator

    Confirmed

    The published claim

    Secondary watch: the reserve data at 14:30 on 8 October will show whether gross reserves fell from 171.2 billion dollars on 25 September to the 167.2 billion dollars Matriks had estimated.

    What happened

    In the CBRT's 8 October data, gross reserves fell to 167.4 billion dollars, only 200 million dollars off the Matriks estimate. Net reserves excluding swaps fell from 39.9 billion dollars to 37.9 billion dollars in a week.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Causal chain

    Falsified

    The published claim

    Main conclusion: if foreign demand softens at a single auction, the dealer share rises from 2.5%, the yield settles above 5.30% and emerging-market premia widen.

    What happened

    The first two links of the chain arrived at the 30-year auction on 8 October: the indirect share fell from 79.5% in September to 72.3%, the dealer share rose to 6.8% and the yield came in at 5.618%. The final link went the other way. The next day, 9 October, Türkiye's CDS fell from 248.43 to 244.07, Mexico's from 94.02 to 91.68, South Africa's from 138.49 to 135.82 and Egypt's from 318.98 to 310.69. Emerging-market premia did not widen; they narrowed.

    Evidence: Investing.com — Turkey CDS 5 Years USD historical data
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: the 6 October fall in Türkiye's CDS was not a local improvement but part of a panel-wide tightening.

    What happened

    Between 6 and 9 October Türkiye's CDS fell from 247.19 to 244.07, or 1.3%. Over the same period Mexico tightened 3.8%, Saudi Arabia 2.0%, Israel 1.1% and South Africa 0.8%. Türkiye moved in step with its emerging-market peers; no idiosyncratic improvement was visible. Europe went the other way, with France rising from 72.07 to 78.58.

    Evidence: Investing.com — Turkey CDS 5 Years USD historical data
  • Scenario

    Still open

    The published claim

    H1 · 50% — The floor is high but financed: the indirect share at the 30-year auction stays above 70%, the 14 October CPI gives the Fed no case for an October hike, and the US 10-year stays in a 5.15–5.40% band; the CBRT makes a small cut on 22 October.

    What happened

    Due 22 October. The first trigger held: the indirect share was 72.3%. The FRED 10-year yield was 5.22% on 8 October, inside the band. The 14 October CPI and the 22 October MPC decision are pending.

  • Scenario

    Still open

    The published claim

    H2 · 30% — Foreign demand softens at one auction: the indirect share at the 30-year or the next 10-year auction approaches 57.2%, the dealer share tops 10% and the US 10-year rises above 5.40%.

    What happened

    The 30-year leg was not triggered: the indirect share held at 72.3%, the dealer share at 6.8%, and the 10-year fell to 5.22% on 8 October. The scenario stays open until the 10-year auction in November.

  • Falsifier

    Still open

    The published claim

    Converging signals: the reading is refuted if, by 31 October, Iraq's crude exports return above 3 million barrels a day, UKMTO reports no new crew injuries inside the Gulf and the Micron Taoyuan mediation ends in a settlement without a strike.

    What happened

    Due 31 October. The interim picture supports the reading: two more outbound tankers were hit in Hormuz on 10 and 11 October, and UKMTO has logged 9 attacks in October. We found no report of a settlement at Micron dated after 8 October.

    Evidence: 11 October 2026 · Second outbound tanker hit in Hormuz in two days

What we learned

Where we were wrong is the final link of the main conclusion. We wrote that softer foreign demand at a single auction would immediately widen emerging-market premia. The softening came at the 30-year, but the next day emerging-market CDS, Türkiye's included, tightened. We drew the link too directly and too fast; there is at least one intermediate variable between the US long end and emerging-market premia, such as the dollar or Fed expectations. On the primary watch we set a two-sided threshold and the result fell squarely in the middle; next time we will also spell out in advance what the middle zone means.

Published on 8 October 2026 · Reviewed on 13 October 2026

Foreign buyers are holding up the floor under global borrowing costs

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 50% — The floor is high but stable: foreign demand at auctions holds above 70%, pricing of a December hike stays around 70%, and the 10-year yield trades in a 5.15–5.45% band; long-dated US yields hold around 5.3%.

    What happened

    At the 8 October 30-year, October's last coupon auction, the indirect share of 72.3% stayed above the threshold. The FRED 10-year was 5.22% on 8 October. The 30-year yield, however, stood at 5.60% that day, well above 5.3%.

    Evidence: 8 October 2026 · Weak demand at US 30-year auction breaks with strong 10-year sale
  • Scenario

    Still open

    The published claim

    H2 · 30% — Foreign demand breaks at one auction: the indirect share at the 3-year or 30-year auction drops below 60%, the dealer share tops 15% and the 10-year yield passes 5.45%.

    What happened

    The trigger did not fire in the October cycle: at the 30-year the indirect share was 72.3% and the dealer share 6.8%, and the 10-year was 5.22% on 8 October. Demand weakened but did not break. The scenario stays open until the November auctions.

    Evidence: 8 October 2026 · Weak demand at US 30-year auction breaks with strong 10-year sale
  • Causal chain

    Still open

    The published claim

    Step 1 (weeks): at an auction where foreign demand stays weak, dealers absorb the unsold paper and the 10-year floor settles above 5.45%.

    What happened

    The chain's breaking condition is holding for now: the indirect share at the 30-year stayed above 70%. Rather than approaching 5.45%, the 10-year yield fell to 5.22% on 8 October. The 10-year auction in November will be decisive.

  • Watched indicator

    Still open

    The published claim

    Threshold: Türkiye's 5-year CDS rising from 247.19 on 6 October to above 275 would show the rise in the US floor passing through to Türkiye's risk premium.

    What happened

    The threshold was not crossed; at 244.07 on 9 October the CDS was below its level on publication day. This step of the chain runs on a horizon of months.

  • Watched indicator

    Still open

    The published claim

    Threshold: the 10-year lira yield rising from 32.75% on 7 October to above 34% would show foreign selling of government bonds feeding into the cost of extended-maturity lira debt.

    What happened

    The Investing.com series shows 35.495% on 12 October, against 32.78% on 9 October. We did not count this jump as a threshold breach. The 2-year series also leapt from 36.35% to 39.53% the same day. Bloomberg HT's compounded 10-year yield, meanwhile, was already at 35.52% before 12 October. This suggests the series switched from simple to compounded yield rather than a market move. Because the threshold was set on simple yield, the verdict stays open until confirmed on the same measure.

    Evidence: Investing.com — Turkey 10-Year Bond Yield historical data

What we learned

Nothing has matured yet, and no verdict has closed. There are two drafting errors. H1's market line said long-dated US yields would hold around 5.3%, whereas the 30-year stood at 5.67% on publication day; the sentence did not say which maturity it meant. The second problem is measurement: we tied the Türkiye 10-year threshold to Investing.com's simple-yield series, which appears to have switched to a compounded basis on 12 October. Rule: every yield threshold will be written with the series name and yield type, and reset if the series type changes.

Published on 8 October 2026 · Reviewed on 13 October 2026

Türkiye's Treasury is borrowing to pay interest that now exceeds principal

Open the report ›
  • Scenario

    Still open

    The published claim

    H2 · 35% — The MPC waits: reserve erosion continues in the 8 and 15 October data, and pressure on USD/TRY pushes the MPC to hold at 37%.

    What happened

    Due 22 October. The 8 October leg held: gross reserves fell by $3.8 billion to $167.4 billion, and net reserves excluding swaps fell to $37.9 billion. The 15 October data and the MPC decision are pending.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Scenario

    Still open

    The published claim

    H1 · 50% — A cut comes: the 22 October MPC lowers the rate below 37%, and the weekly fall in official gross reserves drops below $2 billion.

    What happened

    Due 22 October. The reserve leg did not hold in the 8 October data: the weekly fall of $3.8 billion exceeded $2 billion.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Watched indicator

    Still open

    The published claim

    Threshold: if the 10-year lira yield rises from 32.75% on 7 October above 34.50%, the market is demanding a term premium for debt that grows with interest; H3 strengthens.

    What happened

    Threshold not crossed. In the Investing.com series the 10-year yield was 32.75% on 12 October and 32.80% on 13 October, at the 32.75% level of 7 October when the threshold was set. A reading near 35.5% for 12 October seen in an intermediate source was not confirmed by the series, and our explanation built on it, that "the series switched to compound", was also wrong. There is no sign of demand for a term premium.

    Evidence: Investing.com — Turkey 10-Year Bond Yield Historical Data

What we learned

Due 22 October; no verdict. The reserve data went against H1's trigger and in line with H2's. Measurement risk is present in this report too: the 10-year threshold was tied to the Investing.com simple yield, and the series appears to have moved to a compound level on 12 October. No verdict will be given on this indicator until the threshold is reset.

Published on 8 October 2026 · Reviewed on 13 October 2026

Oil's inventory cushion has worn thin, so shocks now hit prices directly

Open the report ›
  • Causal chain

    Confirmed

    The published claim

    Step 2 (weeks): the Atlantic diesel squeeze feeds into the cost of imported diesel; with the sliding-scale tax mechanism lifted on 1 October, the cost passes to the pump and the 4.80–4.95 lira cut of 6 October is reversed.

    What happened

    The cut was reversed within 4 days, not weeks. On 10 October diesel rose by 6.36 lira a litre, taking the price on Istanbul's European side from 90.20 lira on 6 October to 96.55 lira. We did not separate whether the increase came from the Atlantic squeeze or from Brent's rise.

    Evidence: 10 October 2026 · Turkish diesel jumps over 6 lira as Russia eases its export ban
  • Watched indicator

    Confirmed

    The published claim

    Flagged as noise: the diesel price cut on 6 October does not show that the diesel squeeze has passed.

    What happened

    Four days later, on 10 October, a 6.36 lira increase more than wiped out the 6 October cut. The cut did not last.

    Evidence: 10 October 2026 · Turkish diesel jumps over 6 lira as Russia eases its export ban
  • Scenario

    Still open

    The published claim

    H1 · 55% — The buffer thins and prices stay shock-sensitive: every new headline produces a daily swing of 2–4% in Brent; Brent stays volatile in a 95–108 dollar band, and no diesel price cut in Türkiye lasts.

    What happened

    The interim picture is consistent with H1. In the Investing.com series Brent rose 4.07% on 8 October and 2.48% on 12 October. Closes of 100.20–106.12 dollars sit inside the band. The diesel cut was reversed on 10 October. The scenario has no maturity date; we will check again at the end of October.

    Evidence: Investing.com — Brent Oil Futures historical data
  • Watched indicator

    Still open

    The published claim

    Threshold: an ICE Brent front-month close above 108 dollars, up from 100.20 dollars on 7 October, would show the stock buffer cannot absorb a second shock.

    What happened

    Not yet crossed. The December contract closed at 106.12 dollars on 12 October after touching 107.28 dollars intraday. According to a Finans Gündem report citing Reuters, the settlement price that day was 106.65 dollars.

    Evidence: Finans Gündem — End of day in markets: Brent oil tops 106 dollars
  • Scenario

    Still open

    The published claim

    H2 · 25% — A second shock hits the buffer: Isaias shuts a Gulf Coast refinery for weeks, or a strike on a Gulf pipeline or terminal is confirmed; Brent passes 108 dollars.

    What happened

    The trigger did not fire. Tankers were hit in Hormuz between 9 and 11 October, but no strike on a pipeline or terminal was confirmed. Brent is below 108 dollars.

    Evidence: 11 October 2026 · Second outbound tanker hit in Hormuz in two days
  • Watched indicator

    Still open

    The published claim

    Threshold: US commercial crude stocks falling below 415 million barrels from 424.1 million barrels in the week of 2 October.

    What happened

    EIA data for the week of 9 October have not yet been released.

What we learned

The Türkiye chain held, but at the wrong speed. We wrote that the cut would be reversed within weeks; it was reversed in 4 days. Without a tax buffer, costs reach the pump within days, and we should have written 'days' for the lag on this channel. We did not isolate the cause of the increase, so the Atlantic diesel squeeze mechanism cannot be counted as separately confirmed. The Brent threshold has not yet been tested.

Published on 7 October 2026 · Reviewed on 12 October 2026

Money fleeing funds finds shelter in deposits, leaving non-bank lenders unfunded

Open the brief ›
  • Watched indicator

    Confirmed

    The published claim

    Secondary watch: the 8 October 14:30 reserve release. It will show whether gross reserves fell from $171.2 billion on 25 September to the $167.2 billion estimated by Matriks, and which way net reserves excluding the $40 billion of swaps moved.

    What happened

    In the CBRT's 8 October release for the week of 2 October, gross reserves fell by $3.8 billion to $167.4 billion, only $200 million off the Matriks estimate. Net reserves excluding swaps slipped from $39.9 billion to $37.9 billion. The direction we read held.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Watched indicator

    Still open

    The published claim

    Primary watch: any new MKK default notice from outside the Tera network, and the weekly number of private-sector debt instruments given an issuance quota. If the 13 issues recover above 20, the crisis closes as the problem of a single broker network.

    What happened

    According to an Endeks24 compilation dated 6 October, the MKK published only two default notices in the first six days of October: Tera Yatırım Bankası (5 October) and Adil Varlık Yönetim (6 October), whose broker is the Tera group. No factoring company is on the list. We found no new notice after 6 October and no quota count for the October weeks; the verdict stays open until 22 October.

  • Scenario

    Still open

    The published claim

    H1 · 50% — The leak stays inside the Tera network: no default from outside it by 22 October, weekly issuance above 20, gross reserves above $165 billion on 8 October; the CBRT cuts on 22 October.

    What happened

    Due 22 October. The reserve leg of the trigger held: gross reserves stood at $167.4 billion on 8 October. CDS eased from 255.29 on 5 October to 248.43 on 8 October, and the 2-year yield fell to 36.35% on 9 October; USD/TRY was 49.17 on 9 October. The issuance leg could not be measured.

  • Scenario

    Still open

    The published claim

    H3 · 20% — An external shock squeezes reserves: a confirmed hit on the East-West pipeline, Yanbu or Rabigh, December Brent above $110, and gross reserves below $165 billion in the 8 and 15 October releases.

    What happened

    So far none of the triggers has arrived. Brent futures stood at $104.72 on 9 October. Gross reserves were $167.4 billion on 8 October, above 165. The Houthi attack on 10 October struck Riyadh airport, not the pipeline, killing 12 people. The 15 October release is pending.

    Evidence: 10 October 2026 · Houthi strike kills 12 at Riyadh airport, worst in a decade
  • Watched indicator

    Falsified

    The published claim

    Flagged as noise: a softer-than-expected inflation print did not restore confidence; 5-year CDS sat at 255.69 on 2 October and 255.29 on 5 October. The risk premium is not moving.

    What happened

    The brief relied on the 5 October close. In the Investing.com series, CDS fell to 247.19 on 6 October, the day after the inflation data, a drop of 8.1 basis points in a single session. It closed at 247.80 on 7 October and 248.43 on 8 October. The risk premium did move, and the decline held through 8 October.

    Evidence: Investing.com — Turkey CDS 5 Years USD historical data
  • Watched indicator

    Partly borne out

    The published claim

    Non-bank finance has lost its funding: factoring and leasing companies that borrow through notes are losing their buyers.

    What happened

    According to Bloomberg HT on 7 October, the CMB that same week approved debt issues of 15.15 billion lira by İş Faktoring and 5.073 billion lira by Koç Stellantis Finansman. Non-bank lenders tied to banking groups kept their access to funding. The freeze appears confined to independent and smaller issuers. An approval is not an actual sale.

    Evidence: Bloomberg HT — CMB approves debt issuance by four companies
  • Falsifier

    Still open

    The published claim

    Converging signals: the deadline is 31 October. The reading fails if the Saudi side clarifies daily East-West pipeline flows with written data, the gap between Hormuz estimates narrows below twofold, and control of the Bab el-Mandeb coast is independently confirmed.

    What happened

    Due 31 October. The interim picture supports the reading. On 9 October the Yemeni government said it had taken Bab el-Mandeb; the Houthis denied it, and there is no independent confirmation. In Hormuz, Kpler counted 7 transits on 6 October. No written daily flow data for the East-West pipeline has been published.

    Evidence: 9 October 2026 · Yemen's government claims Bab el-Mandeb as Houthis say they repelled it

What we learned

Our miss was CDS. The brief said the risk premium was not moving, yet CDS fell 8.1 basis points on 6 October and stayed around 248. We built a noise verdict on a single day of data. Change of method: every market gauge in the brief's strip will be rechecked against the latest close on the morning of publication; if the date is more than a day old, it goes into the transparency note. The reserve reading held. The claim that non-bank finance had lost its funding was drawn too broadly: factoring arms of banking groups won issuance approval that same week. The freeze sits with independent issuers.

Published on 4 October 2026 · Reviewed on 12 October 2026

The blockade reroutes oil, and the constraint moves into insurance policies

Open the report ›
  • Causal chain

    Confirmed

    The published claim

    Step 1 (days): because every cargo through Hormuz carries a 3–10% war-risk premium, the physical delivered price stays above the futures contract. Watch: the Dated Brent–front-month spread against a $10 threshold.

    What happened

    The physical price stayed above futures and the spread widened. The EIA Brent spot series printed $135.51 on 2 October, $125.51 on 5 October and $125.44 on 6 October. On the same days ICE December Brent (LCOZ6) closed at $102.25, $100.32 and $100.58, a gap of $33.26, $25.19 and $24.86. Caveat: the EIA spot is not Platts Dated Brent; the report's $19 was calculated from Oilprice's Dated figure.

    Evidence: FRED (EIA) — Crude Oil Prices: Brent - Europe (DCOILBRENTEU)
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Redistribution continues: Iran still cannot load in October, regional exports stay above 17 million barrels a day, and Houthi attacks cause no facility outages.

    What happened

    The first leg held. According to Al Jazeera, Washington says Iran has exported no oil past the blockade line; on 10 October CENTCOM struck a vessel trying to run it. There is no October data for regional exports on the same basis. The third leg is disputed: Bloomberg denied claims that flows stopped at the Khurais station on 4–5 October, and the 10 October Houthi attack targeted Riyadh airport rather than a facility.

    Evidence: 10 October 2026 · US warplane strikes cargo ship trying to run Iran blockade
  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — The constraint turns physical again: Iran steps up tanker attacks in the strait or an Aramco outage is confirmed; volumes fall below pre-war levels and crude rises sharply.

    What happened

    The trigger came right after the report. The week of 28 September–5 October brought at least 12 attacks, the most intense week of the war. A tanker was hit off Qatar on 7 October and off Fujairah on 8 October. Kpler put crude through the strait at 8.5 million barrels a day in the week to 7 October. The price leg did not arrive: ICE December Brent rose from $102.25 on 2 October to $104.72 on 9 October, just 2.4%.

    Evidence: 7 October 2026 · Tanker struck off Qatar as attacks move deeper into the Gulf
  • Scenario

    Still open

    The published claim

    H3 · 15% — Deal and partial opening: Washington and Tehran announce an easing of the blockade or a transit arrangement, and Iranian loadings resume on a limited scale.

    What happened

    The report set no window, so the verdict is open. The interim picture points the other way. On 11 October Rezaei called the US response to the seven conditions inadequate and said Tehran was no longer replying; the blockade continues.

  • Causal chain

    Partly borne out

    The published claim

    Step 2 (weeks): refiners pay the physical price for crude, and the premium passes into petrol and diesel prices. Watch: Türkiye's diesel pump price and the EU diesel average.

    What happened

    The pump price rose. On 10 October diesel went up 6.36 lira a litre to 96.55 lira on Istanbul's European side. Crude's premium was not the only cause; the Russian diesel ban and a global diesel shortage pushed prices the same week. The increase came, but we cannot pin the mechanism on the crude premium alone.

    Evidence: 10 October 2026 · Turkish diesel jumps over 6 lira as Russia eases its export ban
  • Watched indicator

    Still open

    The published claim

    If EU gas storage is below the relaxed 80% target on 1 November, Europe enters winter with a thinner buffer; on 3 October storage was 72.3% full against a seasonal norm of 92.0%.

    What happened

    Due 1 November. Interim data: according to IEEFA, storage stood at 72.4% on 3 October, the lowest for this point in the year since 2011. The series in our data file shows 72.8% on 5 October and 73.5% on 9 October. On 8 October ENTSOG wrote that a cold winter could require demand to be cut by about 7%.

  • Falsifier

    Still open

    The published claim

    If the Dated Brent–futures spread stays below $5 for two weeks, the claim that the constraint sits in the insurance policy weakens. If the EU or UK seizes a Russian-flagged tanker by end-2026, the reading that reflagging offers protection is falsified.

    What happened

    The two-week window for the first falsifier has not closed. At the latest available observation, on 6 October, the gap between EIA spot and December futures was $24.86. The second falsifier runs to end-2026; we found no report of a Russian-flagged tanker being seized between 4 and 12 October.

What we learned

The central thesis held over the first eight days: Iran could not load, and physical barrels stayed $25–33 dearer than paper barrels. But the size of the physical premium depends on the series we use. The report's $19 came from Oilprice's Dated figure; the $25–33 here comes from EIA spot, and the two do not measure the same thing. The contract month is consistent: front-month December (LCOZ6) since 2 October. H2's trigger, which we put at 30%, arrived faster than expected, but the price leg did not. Futures stayed calm while the squeeze built up on the physical and freight side. Rule: the physical premium will be tracked with a single named series.

Published on 7 October 2026 · Reviewed on 12 October 2026

Truce defers tariffs, yet chip and rare-earth suspensions still expire in November

Open the report ›
  • Causal chain

    Falsified

    The published claim

    Evidence item: the three chip bills in the Senate NDAA package cleared committee by 42-0, 36-8 and 42-2; the December conference falls about three weeks before the 10 January deadline.

    What happened

    We attributed the votes to the wrong chamber. According to Roll Call on 23 September, 42-0, 36-8 and 42-2 were the votes cast by the House Foreign Affairs Committee on similar versions of these bills in January, March and April. On the Senate side, only two of the bills are reported as included in the NDAA package; it is not stated whether the third is in it.

    Evidence: Roll Call — AI export controls debate rages as Trump, Xi meet (23 September 2026)
  • Causal chain

    Partly borne out

    The published claim

    Step 3: the December NDAA conference, the 1 January DFARS ban and the 10 January truce deadline pile into the same three weeks.

    What happened

    The NDAA leg of this step has lost its footing. According to the Congressional Research Service update of 1 September, the House passed the NDAA on 22 July by 216-212. In the Senate, the cloture vote on 14 July failed 50-46, the bill has not been rescheduled, and no conference has begun. The December conference assumption rested on the 2025 calendar. The DFARS and 10 January dates still stand; the step falls due in January 2027.

    Evidence: Congressional Research Service (EveryCRSReport) — FY2027 NDAA: Status of Legislative Activity (1 September 2026)
  • Watched indicator

    Still open

    The published claim

    First date to watch, 9 November: the Federal Register notice moving BIS's suspension of the Affiliates Rule to 10 January.

    What happened

    Due 9 November; the verdict is open. In the Federal Register's list of BIS documents, the latest item published between 20 September and 12 October is a recordkeeping notice dated 30 September. There is no notice extending the suspension. The 2 October signal on which the report relied is unchanged as of 12 October.

    Evidence: Federal Register — Bureau of Industry and Security documents since 20 September 2026 (API listing)
  • Watched indicator

    Still open

    The published claim

    China's Ministry of Commerce issuing an official notice moving the rare-earth suspension from 10 November to 10 January.

    What happened

    Due 10 November; the verdict is open. As of 12 October we found no new ministry notice in the sources we checked. Meanwhile, on 9 October Beijing promised the EU a green channel for rare-earth licences. That does not show the suspension being extended for the US, but it confirms Beijing calibrates its controls partner by partner.

    Evidence: 9 October 2026 · Beijing promises the EU a green channel for rare earth licences
  • Scenario

    Still open

    The published claim

    H1 · 55% — Last-minute official extension: BIS and China's Ministry of Commerce both move their suspensions to 10 January 2027 before 9–10 November.

    What happened

    Due 9–10 November; the verdict is open. As of 12 October neither side has published an official text; the probability stands as written.

  • Scenario

    Still open

    The published claim

    H2 · 30% — Partial gap: one side delays its notice, and one of the rules comes back into force for a few weeks under selective licensing.

    What happened

    Due 9–10 November; the verdict is open. No rule came back into force between 5 and 12 October. A broker's guilty plea in the Super Micro case on 8 or 9 October shows Washington also enforces controls through prosecution rather than rules. That neither confirms nor refutes this scenario.

    Evidence: 8 October 2026 · Middleman in Super Micro case pleads guilty to diverting servers to China
  • Watched indicator

    Still open

    The published claim

    Thresholds: USD/CNY above 6.85 and the VIX above 25 would show the November deadline being priced.

    What happened

    Neither threshold was crossed, and USD/CNY moved the other way. In the FRED series USD/CNY was 6.7046 on 5 October and 6.6921 on 9 October; the VIX went from 15.31 on 2 October to 15.41 on 8 October. As of 12 October the market is not pricing a November cliff; the thresholds are being watched until 10 November.

  • Causal chain

    Still open

    The published claim

    Step 2: front-loaded purchases land on tight memory supply; at producers such as Nanya, which meet 50–60% of demand, contract prices rise faster than expected.

    What happened

    The first step, which triggers this one, has not happened; the verdict is open. According to TrendForce on 6 October, Nanya's September revenue rose 0.9% month on month to NT$45.09 billion. Fourth-quarter contract prices have not yet been announced.

    Evidence: TrendForce — Nanya DRAM ASP seen rising 10% in 4Q26, Winbond also sees further price gains (6 October 2026)

What we learned

Our error was the congressional calendar. We attributed the committee votes in Roll Call to the Senate and said all three bills were in the NDAA package. Working from the 18 December signing in 2025, we also made a December conference the third clock in the constraint timetable. In fact the votes belong to a House committee, and the Senate NDAA has been stuck since 14 July. That weakens the report's weakest link: the reading that leverage would pile up at once in the December–January window. The 9–10 November suspension dates still stand. Lesson: no congressional vote goes into a signal or impact item until it is read from the primary record, with chamber and committee named. The probabilities stand as written; the scenarios will be checked again after 10 November.

Published on 7 October 2026 · Reviewed on 12 October 2026

The fund crisis is locking up non-bank credit, not the banks

Open the report ›
  • Causal chain

    Partly borne out

    The published claim

    Step 1 (weeks): factoring, leasing and asset-management companies cannot roll maturing notes with new issues; they have to meet payments through bank loans, asset sales or maturity negotiations.

    What happened

    The step did not play out at companies tied to banking groups. According to Bloomberg HT, the CMB's 7 October bulletin approved issues of 15.15 billion lira by İş Faktoring and 5.073 billion lira by Koç Stellantis Finansman. According to Endeks24, none of the MKK's 43 default notices in 2026 involves a factoring company. We could not find quota data for independent issuers in the October weeks.

    Evidence: Bloomberg HT — CMB approves debt issuance by four companies
  • Causal chain

    Partly borne out

    The published claim

    Step 3 (months): if the CBRT cuts on 22 October, lira yields fall; with the risk premium at 255 basis points, FX demand is met from reserves and net reserves excluding swaps drop below $40 billion.

    What happened

    The effect arrived before the cause. Net reserves excluding swaps fell to $37.9 billion in the week of 2 October, two weeks before the cut. CDS did not stay at 255; it eased to 248.43 on 8 October. Reserves are draining without waiting for the MPC decision; tying the chain to the cut got the sequence wrong.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Scenario

    Still open

    The published claim

    H1 · 50% — The leak stays inside the Tera network and the cut comes: no default from outside the network, weekly issuance above 20, gross reserves holding above $165 billion; the CBRT cuts by 100 basis points on 22 October.

    What happened

    Due 22 October. Interim picture: gross reserves stood at $167.4 billion on 8 October, and we found no new MKK default notice after 6 October. The 2-year yield was 36.35% on 9 October, below the 37% policy rate; USD/TRY was 49.17.

  • Scenario

    Still open

    The published claim

    H2 · 30% — The funding freeze spreads across non-bank finance: at least one factoring or leasing company from outside the Tera network files an MKK default notice; CDS breaks above 270 and USD/TRY tests 50.

    What happened

    Due 22 October. The interim picture shows no trigger: as of 6 October there was no factoring company on the MKK list, CDS was 248.43 on 8 October and USD/TRY 49.17 on 9 October.

  • Scenario

    Still open

    The published claim

    H3 · 20% — Reserves delay the cut: data for 8 October and the following two weeks show gross reserves falling towards $160 billion and FX deposits rising by more than $1 billion a week; the rate stays at 37%.

    What happened

    Due 22 October. The first release, at $167.4 billion, is far from 160, but the decline continues at $3.8 billion. We could not verify the weekly change in FX deposits from a source. The 15 October release is pending.

  • Watched indicator

    Still open

    The published claim

    Threshold: if the 10-year yield rises from 32.82% on 6 October to above 33.5%, the market is reading the cut as an inflation risk.

    What happened

    The series conflict. Investing.com now shows the 9 October close at 32.78%; our 11 October scorecard records the same series at 35.61% on 9 October. According to Borsatek, the domestic benchmark stood at 35.52% on 8 October. The domestic benchmark is above the threshold, the Investing series below it. Because we defined the threshold with two different series, we cannot give a verdict.

  • Falsifier

    Partly borne out

    The published claim

    Condition that breaks the chain: if banks with surplus deposits open credit lines to factoring and leasing companies, the first link breaks; if foreign bond inflows offset reserve losses, the last link breaks.

    What happened

    The last link did not break. In the week of 2 October foreigners put $928.5 million into eurobonds but pulled $484.7 million from government bonds, and reserves fell by $3.8 billion. At the first link we saw not bank credit but issuance approval for banking groups' own factoring arms; the effect is similar, the channel different.

    Evidence: 8 October 2026 · Foreigners exit lira bonds but put 928.5 million dollars into Turkish eurobonds

What we learned

Two errors are plain. The first is sequence. We built a chain that tied net reserves falling below $40 billion to the 22 October cut. In fact net reserves excluding swaps fell to $37.9 billion in the week of 2 October, without waiting for the decision. The reserve drain comes through a channel independent of monetary policy. The second is scope: we treated non-bank finance as a single bloc. Factoring arms of banking groups won issuance approval that same week; the freeze sits with independent issuers. Change of method: non-bank finance thresholds will be written separately for issuers inside and outside banking groups, and yield thresholds will be tied to a single named series.

Published on 7 October 2026 · Reviewed on 12 October 2026

The Gulf's oil buffer has filled, but on a single pipeline

Open the report ›
  • Watched indicator

    Falsified

    The published claim

    Price reference: ICE Brent futures closed 6 October up 0.76% at $101.08, WTI futures up 0.37% at $89.76; the Brent–WTI spread was $11.32. The $110 threshold was set against this close.

    What happened

    The closes in the report do not match the very series we cited. Investing.com's Brent page (LCOZ6, December contract) has 6 October closing up 0.26% at $100.58, and its WTI page up 0.01% at $89.44; the spread is $11.14. The graver error is one of method. Since 1 October Brent's front month has been December and WTI's November. The $11.32 spread compared two different delivery months.

    Evidence: Investing.com — Brent Oil Futures Historical Data (LCOZ6, December 2026)
  • Scenario

    Still open

    The published claim

    H1 · 55% — The single point holds and the premium settles: the East-West pipeline flows above 5 million barrels, no new hit is confirmed; Hormuz's share stays around 60%, Bab el-Mandeb transits in a 30–40 band; Brent around $100.

    What happened

    The thesis runs three months; it is not yet due. Interim picture as of 12 October: we found no confirmed new hit on the East-West pipeline or Yanbu between 7 and 11 October. Houthi attacks turned to Riyadh airport on 10 October, killing 12 people. ICE December Brent closed at $100.20 on 7 October and $104.72 on 9 October. No October data on Hormuz's share has been published on the same basis.

  • Scenario

    Still open

    The published claim

    H2 · 25% — The single point breaks: a confirmed hit on one of the pumping stations, Yanbu or Rabigh; Brent tests $110 and the Yanbu premium approaches 7%.

    What happened

    Not yet due. The interim picture does not point to H2's trigger: we found no confirmed hit on the pipeline or Yanbu on 7–11 October. In the Investing.com series the highest close for ICE December Brent was $104.72 on 9 October. The $105.28 seen on Sunday evening, 11 October, was an intraday value, not a settlement. The $110 threshold was not crossed.

  • Watched indicator

    Partly borne out

    The published claim

    If the share of Gulf barrels passing through Hormuz rises from 60% in September to above 70%, the bypass load eases and the thesis of concentration on a single pipeline weakens.

    What happened

    No October share has been published on the same basis, so the threshold could not be tested directly. Indirect gauges show the share did not rise. Kpler data cited by Al Jazeera put crude through the strait at 8.5 million barrels a day in the week to 7 October, about 40% below normal. The indicator had been chosen without checking its publication schedule.

    Evidence: Al Jazeera — Hormuz attacks intensify as efforts to end war on Iran stall
  • Causal chain

    Still open

    The published claim

    Step 1 (weeks): international insurers cover Saudi-linked Red Sea voyages only at premiums at least ten times higher; the Saudi Re pool tries to fill the gap. Watch: the Yanbu premium (3%) and any capacity statement from the pool.

    What happened

    Since publication we have found no Yanbu quote more recent than the 3% of 25 September, and no capacity statement from the Saudi Re pool. The step was written with a lag of weeks; there is no measurable data in the first five days.

  • Causal chain

    Still open

    The published claim

    Step 2 (weeks): the premium is added to the delivered price of barrels loading at Yanbu, and Bab el-Mandeb transits stay low. Watch: Aramco's December Mediterranean differential and daily Bab el-Mandeb transits.

    What happened

    Aramco will announce December prices in early November. IMF PortWatch's Bab el-Mandeb series ends on 4 October at 34 transits; between 27 September and 4 October the daily count ranged from 22 to 37. The series does not yet cover the days after publication.

  • Causal chain

    Still open

    The published claim

    Step 3 (months): Mediterranean importers such as Türkiye pay a higher energy bill as volumes fall. Watch: the September–October energy import bill and crude import volumes.

    What happened

    In the Trade Ministry's preliminary September bulletin, total imports rose 5.9% to $31.2 billion. Export cover of imports was 83.2%, and 100.3% excluding energy. The bulletin gives neither the energy bill nor crude volumes; TÜİK will publish them on 30 October.

What we learned

Our miss was the price series, again. The $101.08 and $89.76 we gave as 6 October closes were not settlement values in the series we cited; most likely they were intraday figures. We also built the Brent–WTI spread between December Brent and November WTI. This is the third repeat of the contract-month error recorded in the 8 and 11 October scorecards. Change of method: spreads will be calculated only between two contracts with the same delivery month, and closes will be taken after settlement from a series identified by name and contract code. The scenarios run three months; in the first five days the single point did not break, and Houthi attacks targeted the capital's airport rather than the pipeline.

Published on 7 October 2026 · Reviewed on 12 October 2026

AI has opened an inflation channel the Fed's rates cannot reach

Open the report ›
  • Watched indicator

    Confirmed

    The published claim

    The September minutes, due at 21:00 TRT on 7 October, will show how often AI demand was cited as an inflation risk behind the 12–0 hike; the thesis rests on the committee naming this channel.

    What happened

    The minutes came out on 7 October and cited AI investment as a source of inflation in several places. Participants noted that AI investment was adding to inflationary pressure, that prices of AI-linked consumer technology were rising, and that investment could push aggregate demand ahead of supply over the medium term. The same text notes that AI could raise productivity. It never uses the words chip or memory: it ties the channel to investment demand in general, not to memory prices as we did.

    Evidence: Federal Reserve — Minutes of the FOMC, September 15–16, 2026
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Hold in October, December door open: the minutes make further tightening conditional, September core CPI comes in near expectations, the Fed holds at 3.75–4.00% on 27–28 October; the 10-year stays in a 5.2–5.4% range.

    What happened

    Only one of the three triggers has arrived. In the minutes most participants judged one more hike by year-end likely to be appropriate, with no concrete signal for October. FedWatch puts the odds of an October hike at 17% and December at about 70%. The 10-year yield stayed in range, closing at 5.28%, 5.22% and 5.24% on 7–9 October. The 14 October CPI and the 27–28 October decision are still to come; the scenario itself is open.

    Evidence: 7 October 2026 · Fed minutes point to one more rate rise before year-end
  • Scenario

    Still open

    The published claim

    H2 · 30% — The chip channel reaches prices: computer and electronics items accelerate in the 14 October CPI, the 2-year yield breaks above 5%, and the Fed makes a December hike its base case.

    What happened

    September CPI is due on 14 October. The 2-year yield was 4.80% on 9 October, 20 basis points below the 5% threshold.

    Evidence: U.S. Treasury — Daily Treasury Par Yield Curve Rates, October 2026
  • Scenario

    Still open

    The published claim

    H3 · 20% — Demand cools and the door closes: labour-market weakness deepens, AI investment is cut back, and the 1-year yield falls below 4.00%.

    What happened

    The 1-year yield closed at 4.42%, 4.44% and 4.47% on 7–9 October, nowhere near 4.00%. On 8 October Waller played down September's weak payrolls and said hikes need not come at consecutive meetings. October employment data are not yet out.

    Evidence: 8 October 2026 · Fed's Waller says hikes need not come at consecutive meetings
  • Causal chain

    Still open

    The published claim

    Link 2: seeing a rate-insensitive source of demand, the Fed moves a December hike into its base case; 1- and 2-year yields rise and the dollar strengthens.

    What happened

    The market prices a December hike at about 70%, yet the front end has not moved. The 2-year yield went from 4.79% on 6 October to 4.80% on 9 October, and the 1-year from 4.46% to 4.47%: one basis point. The link's lag is weeks; the 27–28 October FOMC is the test.

    Evidence: U.S. Treasury — Daily Treasury Par Yield Curve Rates, October 2026
  • Watched indicator

    Still open

    The published claim

    Thresholds: a close above 5.35% on the 10-year yield, above 5.70% on the 30-year, and above 33.50% on Türkiye's 10-year yield.

    What happened

    None was crossed. In Treasury data the 10-year's highest close on 7–9 October was 5.28%, and the 30-year's 5.67% on 7 October, 3 basis points short of its threshold. Türkiye's 10-year yield was 32.78% on 9 October. The thresholds carry no deadline, so the verdict is open.

    Evidence: U.S. Treasury — Daily Treasury Par Yield Curve Rates, October 2026
  • Falsifier

    Still open

    The published claim

    What would falsify our reading: core goods prices easing in the 14 October CPI while the 1-year yield falls below 4.00%.

    What happened

    CPI is due on 14 October; the 1-year yield was 4.47% on 9 October, 47 basis points above the threshold.

What we learned

Five days on, the report's central claim held: the Fed minutes explicitly listed AI investment among the sources of inflationary pressure. But we built the mechanism in the wrong place. We tied the channel to memory and chip prices. The minutes mention neither; they speak of investment demand in general and of technology product prices, and they cite productivity gains as an offset. The chain's second link has not moved either: despite December pricing, the 2-year yield changed by one basis point in three days. Lesson: when anticipating an institution's internal debate, the yardstick should be the language the institution uses, not the sub-channel we chose. The CPI and FOMC verdicts will be checked again on 29 October.

Published on 6 October 2026 · Reviewed on 11 October 2026

One oil shock writes Moscow a currency windfall and Ankara a bill

Open the brief ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — A diffuse shock and a thinning buffer: Brent stays in a $95–105 range, attacks in Hormuz continue but transits rise, and Gulf product exports in October stay at 60–75% of pre-war levels. The diesel premium stays high in the Mediterranean and Europe, and US long-dated yields hold in a 5.2–5.4% range.

    What happened

    The price legs held; the flow leg did not. December Brent sat at the top of the range at $104.28 on 8 October and $104.72 on 9 October, and the US 10-year yield was inside its band at 5.24% on 9 October. Attacks continued but transits did not rise: Kpler counted only 7 commodity vessels through the strait on 6 October, the lowest since 23 July. Nor did the diesel premium stay high. European gasoil fell 3.44% on 9 October after the US licensed Russian diesel. October data on Gulf product exports are not yet available.

    Evidence: 8 October 2026 · Hormuz transits fall to a two-month low
  • Scenario

    Still open

    The published claim

    H2 · 25% — The product gap closes: Gulf refined product exports exceed 80% of pre-war levels in October and Russia does not extend its diesel export ban on 31 October. The diesel spread narrows and Türkiye's CDS falls below 250 bp.

    What happened

    Not yet due; Russia's ban decision comes on 31 October. Two of the outcomes arrived before the trigger did. Türkiye's 5-year CDS fell below 250, at 247.19 basis points on 6 October and 248.43 on 8 October. The diesel spread narrowed, but because of Washington rather than the Gulf: NYMEX diesel futures fell 2.91% after General License 135 on 9 October. Our scenario set did not include this policy channel.

    Evidence: 9 October 2026 · US grants Russian diesel a temporary licence until April 2027
  • Scenario

    Still open

    The published claim

    H3 · 20% — A second shock hits the buffer: a verified supply disruption on the Yanbu line, in Bab el-Mandeb or in the Black Sea pushes Brent above $110; the US 10-year yield could rise above 5.35%.

    What happened

    The brief set no deadline; as of 11 October the trigger had not been hit. December Brent posted its highest close of the week at $104.72 on 9 October and did not reach $110. The risk grew but did not turn into a supply disruption: on 9 October the Revolutionary Guards said they would also strike ships outside the strait. The 10-year yield closed 9 October at 5.24%.

    Evidence: 9 October 2026 · Revolutionary Guards threaten ships beyond the Strait of Hormuz
  • Watched indicator

    Confirmed

    The published claim

    Primary watch item: CBRT reserve data on 8 October. If Matriks's $167.2 billion estimate of gross reserves is confirmed and net reserves excluding swaps fall below $39.9 billion, the decline stretches into a seventh week, and reserves become the deciding variable at the 22 October MPC.

    What happened

    Both measurable conditions were met. In data for the week of 2 October, published on 8 October, the CBRT's official reserves fell by $3.8 billion to $167.4 billion, only $200 million above the Matriks estimate. Net reserves excluding swaps fell from $39.9 billion to $37.9 billion. Whether reserves prove decisive at the MPC remains open until the 22 October decision.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Watched indicator

    Partly borne out

    The published claim

    Secondary watch item: the RBI decision on 7 October and the start of Russia's daily FX purchases of 12.7 billion roubles would show, on the same day, how the shock is shared between importer and exporter balance sheets.

    What happened

    The importer leg was confirmed. On 7 October the RBI raised its repo rate by 25 basis points to 5.50%, its first hike since February 2023. Despite the hike the rupee weakened to 96.88 on 8 October, and reserves fell to $734.6 billion as of 2 October. We could not verify any data this week showing that Russia's purchases had actually begun; the exporter leg remained unproven.

    Evidence: 8 October 2026 · Rupee slides to 96.88 per dollar despite rate hike
  • Watched indicator

    Confirmed

    The published claim

    Calendar and Module C: the $22 billion US 30-year auction on 8 October is the day demand at the long end is tested; negative bias on US long-dated Treasuries in the base case.

    What happened

    The auction was weak. On 8 October the Treasury sold $22 billion of 30-year bonds at 5.618%, and the bid-to-cover ratio fell to 2.54. The indirect bidder share fell to about 72.3% from 80.3% at the previous day's 10-year auction, and the dealer share rose to about 6.8%.

    Evidence: 8 October 2026 · Weak demand at US 30-year auction breaks with strong 10-year sale
  • Falsifier

    Still open

    The published claim

    The converging signal (maritime security being handed to coastal states) weakens if, by 28 October, the US restarts multinational escorts in the Red Sea or Hormuz, and if no Turkish or Pakistani troop deployment on Saudi soil under the Mecca pact is officially confirmed.

    What happened

    Due on 28 October. As of 11 October multinational escorts had not resumed; on 9 October US Central Command said only that it had cleared mines from the main transit lanes. Middle East Eye reported on 6 October that Ankara is preparing additional air defence and a technical team for Saudi Arabia, but a broad deployment awaits parliamentary approval and there is no official confirmation.

What we learned

We read the price band right and the flows wrong. Brent and the US 10-year yield stayed within their H1 ranges, but instead of rising, Hormuz transits fell to a two-month low of 7 vessels on 6 October. The real miss was the diesel premium. We wrote that the diesel spread would narrow only through Gulf exports or Russia's 31 October decision. On 9 October gasoil fell 3.44% on a US sanctions licence, a policy channel we had not modelled at all. Türkiye's CDS also slipped back below 250 even as reserves fell to $37.9 billion. Change of method: in energy and macro scenarios, policy levers such as sanctions licences and stock releases will be written as a separate trigger.

Published on 4 October 2026 · Reviewed on 11 October 2026

Chip controls hold at the sale, then fail at end-use checks

Open the report ›
  • Causal chain

    Falsified

    The published claim

    Noise item: the reading that Malaysia's permit rule of 14 July 2025 solved the transshipment problem. As evidence, we wrote that the scheme ran through Malaysia until 12 August 2026, meaning it kept operating after the rule took effect.

    What happened

    Our evidence rested on the wrong date. According to the US Department of Justice statement of 1 October 2026, the scheme covered 2023–2024, and Earthmade received payments between January and October 2024. The case therefore ended before Malaysia's rule took effect in July 2025. The department also names Singapore alongside Malaysia as a transit point. This case does not show the rule failing; there is still no data on its effect.

    Evidence: US Department of Justice — California man arrested for smuggling more than $300 million in export-controlled servers to China
  • Causal chain

    Confirmed

    The published claim

    Start of the chain: on 1 October charges were filed alleging that controlled Nvidia servers flowed to China through two Malaysian intermediaries for at least 10 months.

    What happened

    The Department of Justice said Earthmade received more than $176 million from two Malaysian logistics companies, and that the payments came between January and October 2024, roughly 10 months. A single order of 27 servers was paid at about $7.61 million. The defendant faces three charges; the two most serious carry up to 20 years in prison each.

    Evidence: US Department of Justice — California man arrested for smuggling more than $300 million in export-controlled servers to China
  • Scenario

    Still open

    The published claim

    H1 · 50% — Patch upon patch: location verification is not enacted by 10 January, the remote access bill waits in the Senate, and BIS's budget rises partially.

    What happened

    Due on 10 January 2027; verdict open. As of 11 October the sources we reviewed show no vote in the Senate Banking Committee on the remote access bill (S. 3519, and H.R. 2683, which has passed the House). The conference text of the 2027 defence authorisation bill, which includes location verification, has not been published. The level at which BIS's fiscal 2027 appropriation has been applied since 1 October could not be verified.

  • Scenario

    Still open

    The published claim

    H2 · 30% — Enforcement moves to access: the defence bill includes location verification, the remote access bill passes the Senate, and BIS's overseas staff rises to 65.

    What happened

    Due on 10 January 2027; verdict open. None of the three triggers had been met as of 11 October, but none has been ruled out either. The latest information that the Senate bill sits in committee and that BIS's overseas staff remains at 25 rests on testimony from July 2026.

  • Scenario

    Still open

    The published claim

    H3 · 20% — Controls enter the bargain: in the January package Washington licenses more chip types, and the congressional bills stall on executive objections.

    What happened

    Due on 10 January 2027; verdict open. No licence for a new chip type was announced between 4 and 11 October. This week Beijing gave the EU a green channel on rare earths and won dual-use licence priority in return. That is a precedent showing technology controls can become a bargaining chip, but it is not evidence for the US package.

    Evidence: 9 October 2026 · Beijing promises the EU a green channel for rare earth licences
  • Watched indicator

    Still open

    The published claim

    Thresholds: USD/CNY above 6.90 would show chip issues hardening in the 10 January talks; below 6.65 would signal expectations of wider chip access.

    What happened

    Neither threshold was crossed. FRED data show USD/CNY at 6.7046 on 2 October and 6.6921 on 9 October; it moved 0.04 closer to the lower threshold but stayed above 6.65. The thresholds will be checked again up to 10 January 2027.

  • Causal chain

    Still open

    The published claim

    Step 2: as physical transshipment grows costlier, Chinese buyers turn to cloud rental and distillation; the item to watch is the Senate Banking Committee vote on the remote access bill.

    What happened

    This is not a step that can be measured in one week. The committee vote had not taken place as of 11 October. This week CrowdStrike disclosed that a Chinese-speaking operator used at least four commercial language models together in attacks on Korean banks. That showed again that model access sits outside export controls; it does not confirm the step, but it does not weaken it either.

    Evidence: 8 October 2026 · CrowdStrike links Korean bank hacks to a China-based AI agent

What we learned

Where we were wrong was a date. We wrote that Malaysia's rule had failed, based on the 12 August 2026 end date given by The Register. Bloomberg gave the end date as August 2024, and we flagged the conflict in the report, yet still rested the noise item on only one of them. The Department of Justice's primary statement gives 2023–2024, so we have no evidence on the effect of Malaysia's rule, positive or negative. The lesson: when sources conflict on a date, no signal or noise evidence is written before the primary document (the indictment or the department's statement) has been read. The scenarios will be checked again on 10 January 2027; the probabilities stand as written.

Published on 30 September 2026 · Reviewed on 11 October 2026

A data fog settles on oil as two major supply sources blur

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — The fog persists and headlines set the price: US–Iran talks drag on without result, flows cannot be independently verified, December Brent swings on headlines within a $97.83–105.29 range, and backwardation holds.

    What happened

    The talks produced nothing: Putin passed Iran's position to Trump on 9 October, and no deal was announced. Prices moved on headlines. In the Investing.com series Brent rose 4.37% on 1 October on the aircraft-carrier report and 4.07% on 8 October on the Hormuz attacks. Closes between 1 and 9 October stayed inside the range at $100.20–104.72, but an intraday $105.92 was seen on 8 October. The swings were not sideways; the price climbed to the top of the range in eight trading days.

    Evidence: 10 October 2026 · Putin relays Tehran's terms for ending the war to Trump
  • Scenario

    Falsified

    The published claim

    H2 · 25% — Flows are verified and the premium unwinds: Washington responds positively to Iran's proposal, visible Hormuz transits rise clearly above 24 vessels, and December Brent falls below $97.83.

    What happened

    None of the three legs arrived within the report's own 10-day window. Washington did not respond positively; on 8 October Trump said he would not strike Iran before 3 November, but the blockade continued. Kpler counted visible transits falling to 7 vessels on 6 October and 10 on 7 October. Brent's lowest close was $100.20 on 7 October; it came nowhere near $97.83.

    Evidence: Alhurra — Tanker attacks spread inside Gulf as Hormuz traffic hits 2-month low
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — The fog turns into bad news: talks collapse or a new attack occurs, weekly Kpler flows stay below 33.7 million barrels, December Brent breaks above $105.29, and backwardation steepens.

    What happened

    The attack leg arrived: Reuters counted at least 12 tanker attacks in the week of 28 September–5 October, the most intense week of the war. The flow leg went the other way. In its 6 October note Kpler put crude through the strait at no less than 10.1 million barrels a day, more than 70 million barrels a week. The price leg reached the threshold but did not close above it: the highest close was $104.72 on 9 October, the intraday peak $105.92 on 8 October.

    Evidence: Alhurra — Tanker attacks spread inside Gulf as Hormuz traffic hits 2-month low
  • Causal chain

    Partly borne out

    The published claim

    Link 1: because flows cannot be measured, prices react to daily headlines; the contract roll on 1 October adds to the noise by making the headline price look about $7 lower.

    What happened

    The headline reaction was confirmed: two daily jumps of more than 4% came on 1 and 8 October. The contract-roll forecast was built wrong. The Investing.com series shows no $7 drop on 1 October; the price rose from $98.03 to $102.31. The same series closes 29 September at $97.09, whereas the report had that day's close at $102.35. The two sources we used were showing different contract months, and our 28 September reference of $97.83 is $99.78 in this series.

    Evidence: Investing.com — Brent Oil Futures historical data
  • Causal chain

    Confirmed

    The published claim

    Link 2: once the 10-day window of the Russian decree runs out, diesel losses are estimated from ship tracking rather than official data, and an uncertainty premium is added to product prices.

    What happened

    As of 10 October the figure used for Russian diesel supply comes from ship tracking: Windward and Vortexa put diesel and gasoil exports at 190,000 barrels a day in the third quarter. The Russian government announced only the 500,000-tonne export tranche and released no output data; Novak's claim that output is sufficient cannot be verified. European diesel futures rose 6% on 8 October, and US diesel futures fell 4% on 9 October.

    Evidence: 10 October 2026 · Russia partly lifts diesel export ban with a 500,000-tonne tranche
  • Causal chain

    Still open

    The published claim

    Link 3: in Brent-indexed Türkiye the bill is set at the prompt price, so the $7.46 curve spread does not show up as a discount; Russian diesel purchases can be tracked only through Turkish customs data.

    What happened

    The lag for this link was set in months. Türkiye's October energy import data have not yet been published; the verdict stays open until the foreign trade release in mid-November.

  • Falsifier

    Partly borne out

    The published claim

    Condition that breaks the chain: if visible Hormuz transits rise clearly above 24 vessels and Kpler flows approach the official 13 million barrels, the chain breaks at the first link.

    What happened

    The flow leg largely materialised soon after publication. In Kpler's 6 October note flows fell 27% in a week to 10.1 million barrels. By our calculation the previous week's wartime peak was about 13.8 million barrels a day, above the official figure. The transit leg did not hold: Kpler counted more than 20 transits a day on 4–5 October and 7 on 6 October; US Navy data showed 32 a day in the 72 hours before 4 October.

    Evidence: Alhurra — Tanker attacks spread inside Gulf as Hormuz traffic hits 2-month low
  • Watched indicator

    Partly borne out

    The published claim

    Three indicators for 10 days: whether December Brent falls below $97.83, whether the share of dark transits in Windward data exceeds 58%, and which series Russian agencies withdraw from publication.

    What happened

    Only the price indicator could be read, and it showed no unwinding of the premium: the lowest close in 1–9 October was $100.20. We could not access Windward's share of dark transits for this check. Nor could we find a public list of the series withdrawn by Russian agencies. Two of the three indicators could not be checked, which shows their accessibility was not tested before publication.

    Evidence: Investing.com — Brent Oil Futures historical data

What we learned

The report's central thesis, that flows from two major supply sources could not be measured, held for Russia and half held for Hormuz. Kpler flows approached the official 13 million barrels within a week; the 1.8- to 3-fold gap was not a lasting fog but partly a lag in tracking data. We gave H2 25%, and none of its three legs arrived in 10 days. Prices did not stay in H1's sideways range but climbed towards the H3 threshold. The worst error was again in the threshold series. The $97.83 reference and the forecast for the 1 October contract roll were built from two sources with different contract months and did not match the Investing.com series. We had recorded this error in the 8 October scorecard as well. The rule: every price threshold is tied to a single named series and contract month, and every watch indicator is tested for accessibility before publication.

Published on 26 September 2026 · Reviewed on 11 October 2026

Türkiye's risk premium comes from within: CDS passes 250 as an 18.3-billion-dollar fund liquidation and a MASAK probe test the intermediation chain

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — The governance premium becomes entrenched: liquidation moves slowly on its six-month timetable, the EMİS investigation widens, and no new fund is closed. CDS stays in a 240–265 range and does not return to the 220s of early September; the lira is held flat with reserves.

    What happened

    The outcome held; part of the trigger did not. According to Investing.com, CDS moved from 248.08 on 25 September to 248.43 on 8 October, and every close in between stayed inside the range, between 246.40 (30 September) and 255.69 (2 October). USD/TRY was flat at 49.27 on 9 October. But liquidation did not move slowly. On 1 October the CMB, the capital markets regulator, opened an interim payment of 1 million lira per fund, and on 8 October some 43,000 investors in 17 funds were paid.

    Evidence: Investing.com — Turkey CDS 5 Years USD historical data
  • Scenario

    Falsified

    The published claim

    H2 · 25% — Contagion: redemption curbs appear at other asset managers or arrests reach brokerages. The domestic and external channels compound, CDS breaks above 270 and USD/TRY tests the 50 threshold.

    What happened

    The trigger came; the outcome did not. Tera executives were detained on 27 September, the investigation reached factoring on 28 September, and on 30 September the BDDK, the banking regulator, transferred the Tera, Destek and Hedef investment banks and two factoring companies to the TMSF deposit insurance fund. A roughly 6.05 billion lira commercial paper of Tera Yatırım Bankası went unpaid at maturity. Even so, CDS peaked at 255.69 on 2 October and came nowhere near 270; USD/TRY stayed at 49.27.

    Evidence: 30 September 2026 · Türkiye seizes three investment banks caught up in the fund probe
  • Scenario

    Falsified

    The published claim

    H3 · 25% — Institutional reassurance: liquidation payments start ahead of schedule, the CMB and MASAK make concrete statements on data integrity, and global yields ease. CDS returns below 235, close to its 18 September level.

    What happened

    The strongest part of the trigger came: despite the six-month liquidation timetable, the first payments were made in 17 funds on 8 October. We found no official reassurance from MASAK, the financial crimes unit, on data integrity. CDS did not fall to 235; the period's lowest close was 246.40 on 30 September, and the 8 October close was 248.43.

    Evidence: 8 October 2026 · First payout in Türkiye's fund liquidation reaches 43,000 investors
  • Watched indicator

    Confirmed

    The published claim

    The roughly 300-basis-point jump in the 10-year yield on 25 September could not be verified; it was treated as noise and the report did not rest its thesis on it. Threshold: a sustained move above 36.

    What happened

    Treating it as noise proved right. The Investing.com series in our data file later showed 25 September at 32.77%; the same series jumped again in a single day on 9 October, from 32.84% to 35.61%. According to Borsatek, the domestic 10-year benchmark had already closed 8 October at 35.52%. The jump was not a market move but a gap between two series; the 36 threshold was not crossed in either.

    Evidence: Borsatek — Latest on BIST and global markets, 9 October 2026
  • Causal chain

    Confirmed

    The published claim

    Step 1 (days): holders of locked savings and other fund investors put liquidity first and move from portfolio funds into deposits or foreign currency.

    What happened

    CBRT Governor Karahan said on 6 October that money leaving funds had largely gone into deposits, and that the share of lira deposits had risen above 61%. According to the MKK, the central securities depository, foreigners' fund holdings fell from $11.1 billion at the end of August to $5 billion at the end of September. The shift went mainly into lira deposits, not foreign currency.

    Evidence: 6 October 2026 · Karahan says money leaving funds has largely returned to deposits
  • Causal chain

    Partly borne out

    The published claim

    Step 2 (weeks): redemption pressure turns into selling of equities and private-sector debt; demand weakens for bank and corporate bond issues. Watch: the BIST 100 against its 24 September close of 12,888.

    What happened

    According to Bloomberg HT, the BIST 100 stood at 12,266 on 9 October, about 4.8% below 12,888. A commercial paper issued by Tera Yatırım Bankası on 6 April with a 182-day tenor went unpaid at maturity. We could not find a dated series showing private-sector bond yields over Treasury bills, so the measurement leg of this step could not be verified.

    Evidence: Türkiye Today — Tera's bank fails to pay $123M bond after fund defaults roil Turkish market
  • Causal chain

    Still open

    The published claim

    Step 3 (months): governance uncertainty keeps the premium foreigners demand high; the cost of rolling external debt for the Treasury and banks stays high even if global rates ease. Watch: CDS staying above 250 and the yield on the next eurobond.

    What happened

    The lag is months; it has not yet run out. CDS was above 250 between 1 and 5 October and fell into a 247–248 band after 6 October; the 8 October close was 248.43. The Treasury's eurobonds drew $928.5 million of net foreign inflows in the week of 2 October. The verdict will be given once the yield on the next eurobond issue is known.

  • Falsifier

    Partly borne out

    The published claim

    Condition that breaks the chain: liquidation payments starting ahead of schedule, no redemption curbs at other funds, and official reassurance from MASAK on data integrity would cut the chain at the first step.

    What happened

    The first part of the condition was met: payments began on 8 October, well ahead of the six-month timetable. We found no reassurance from MASAK. The chain was not cut at the first step, because the shift from funds into deposits had already happened before payments began, prior to 6 October. We had tied the falsifier to a link in the chain that had already played out.

    Evidence: 8 October 2026 · First payout in Türkiye's fund liquidation reaches 43,000 investors

What we learned

Our clearest miss is H2. Contagion came exactly along the path we described: the investigation reached brokerages and factoring, three investment banks passed to the TMSF, and a commercial paper went unpaid. But CDS went no higher than 255.69, not 270. In H3 the trigger also came, and CDS did not fall. In both scenarios CDS did not react to developments in the domestic channel. The error was to measure domestic-channel risk with an external market gauge. Foreigners priced the risk by exiting fund holdings and lira derivatives, while CDS moved sideways with global yields. Change of method: in crisis scenarios of domestic origin, thresholds will be tied not to CDS but to foreign fund and derivative positions and to the corporate bond premium. Treating the jump in the 10-year yield as noise was right; the rule of making the domestic benchmark the primary series came out of this.

Published on 24 September 2026 · Reviewed on 10 October 2026

Hormuz price decouples from flow: Brent lost $9.50 in 5 sessions, regained $4.06 on 23 September; 3 ships in the strait, US stocks up 2.97 million barrels

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Price and flows stay decoupled: the list of conditions stays at seven, visible transits remain below 15 and the East-West pipeline ramps up slowly. Brent futures stay in a $97–105 band with high volatility, and the Brent–WTI spread stays wide.

    What happened

    The band mostly held but was breached twice. Brent futures closed at $106.60 on 24 September and $105.28 on 28 September; between 29 September and 8 October they stayed within $100.20–104.13. The spread leg was confirmed: the Brent–WTI gap widened to $11.91 on 25 September. The leg on a slow pipeline ramp-up proved wrong: the East-West pipeline was back to 5.8 million barrels a day on 6 October.

    Evidence: 25 September 2026 · Brent–WTI spread widens from $7.97 to $11.91 in a week: WTI ends the week lower at $92.41, Brent flat at $104.32
  • Scenario

    Partly borne out

    The published claim

    H2 · 25% — Bypass and diplomacy bring relief together: the East-West pipeline approaches 4 million barrels a day and the US and Iran agree on a written timetable. Hormuz transits settle above 15, and Brent futures fall durably below $99.25.

    What happened

    The bypass leg beat expectations: the Saudi energy minister said on 6 October that the pipeline was back at 5.8 million barrels a day. The diplomacy leg did not arrive: Trump rejected Iran's seven-day plan on 26 September. Nor did the price leg: Brent futures stood at $104.13 on 8 October and never closed below $99.25. Flows recovered; prices did not ease.

    Evidence: 6 October 2026 · Saudi East-West pipeline returns to 5.8 million barrels despite attack claim
  • Scenario

    Falsified

    The published claim

    H3 · 20% — The bypass is hit too: the East-West pipeline or Yanbu halts again, or Bab el-Mandeb transits fall below 20; Brent futures return above $105.

    What happened

    There were attempted attacks but no halt. The Houthis fired six ballistic missiles at Yanbu on 24 September, which the coalition said it intercepted. Following the 5 October claim about Khurais, Bloomberg reported that flows continued. According to IMF PortWatch, daily Bab el-Mandeb transits stayed between 22 and 37 from 22 September to 4 October and never fell below 20. Brent's move above $105 lasted two sessions and came without the H3 trigger.

    Evidence: 24 September 2026 · Houthis fire 6 ballistic missiles at Yanbu and Taif; with the Red Sea end of the East-West pipeline targeted, Brent closes up 3.41% at $106.60
  • Watched indicator

    Falsified

    The published claim

    Threshold: a sustained rise in daily Hormuz transits above about 15 ships, the 10-day average, would be the first sign of the physical reopening that prices were waiting for.

    What happened

    A physical reopening did come, but not through this indicator. Transits rose to 24 on 27 September and above 20 on 4–5 October, then fell to 7 on 6 October; nothing stuck. Meanwhile, Kpler data show outflows via the Gulf of Oman and the Red Sea rose to 6.7 million barrels a day, keeping Middle East crude exports at pre-war levels. We treated the ship count in the strait as a proxy for flows; flows recovered by bypassing the strait.

    Evidence: 8 October 2026 · Hormuz transits fall to a two-month low
  • Watched indicator

    Confirmed

    The published claim

    Threshold: Brent futures above $105 would mean the five-session decline had been fully erased and the reopening had been priced out.

    What happened

    The threshold was crossed the day after publication: Brent futures closed at $106.60 on 24 September. Our reading also held: on 26 September Trump rejected Iran's seven-day plan and hopes of a reopening came off the table. It did not last; from 29 September to 8 October the price stayed below $105.

    Evidence: 24 September 2026 · Houthis fire 6 ballistic missiles at Yanbu and Taif; with the Red Sea end of the East-West pipeline targeted, Brent closes up 3.41% at $106.60
  • Watched indicator

    Falsified

    The published claim

    Threshold: US commercial crude stocks falling back below 423.4 million barrels would confirm that the 2.97 million-barrel build was a one-week export anomaly.

    What happened

    Stocks did not fall back below the threshold. EIA data put commercial crude stocks at 426.398 million barrels in the week of 18 September, 427.320 million in the week of 25 September and 424.134 million in the week of 2 October. The build extended by a further 0.9 million barrels the following week. Three weeks of data did not support the 'one-week anomaly' reading. The thesis that the build stayed local was, however, confirmed by the spread.

    Evidence: EIA — Weekly U.S. Ending Stocks excluding SPR of Crude Oil (WCESTUS1)
  • Causal chain

    Falsified

    The published claim

    If the East-West pipeline cannot reach 4 million barrels a day within weeks, Saudi exports stay dependent on Hormuz and Yanbu cargoes get stuck at Bab el-Mandeb.

    What happened

    The condition did not materialise, and the chain stopped at this link. Yanbu resumed loadings on 28 September after a 17-day gap; the pipeline reached 5.8 million barrels a day on 6 October. On 6 October the EIA wrote that bypass routes had restored Gulf flows to 90% of pre-war levels. We overestimated how long the pipeline would take to recover.

    Evidence: 6 October 2026 · Draining stocks push the EIA's year-end Brent forecast to $105
  • Falsifier

    Falsified

    The published claim

    Condition that breaks the chain: if Hormuz transits rise durably above 15 and the East-West pipeline approaches 4 million barrels a day, physical flows catch up with expectations and Brent falls durably below $99.25.

    What happened

    The pipeline leg happened, the transit leg did not, and the price outcome never came. Despite the recovery in flows, Brent stood at $104.13 on 8 October. On 6 October the EIA raised its fourth-quarter Brent forecast to $105 and pointed to inventories: global stocks drew by 1.9 million barrels a day in the third quarter. We tied price to flows; price was tied to depleted inventories.

    Evidence: 6 October 2026 · Draining stocks push the EIA's year-end Brent forecast to $105

What we learned

We were wrong in two places. The first was the indicator: we treated the ship count in Hormuz as a proxy for physical flows, yet exports bypassed the strait and returned to pre-war levels while the count swung between 7 and 24. The second was what drives the price: we wrote that prices would fall once flows recovered, but Brent held at $104 because inventories drawn down in the third quarter kept price independent of flows. We also overestimated by weeks how long the East-West pipeline would take to recover. Change of method: falsifiers in supply reports will be tied jointly to total regional exports and inventory changes, not to a single transit count.

Published on 29 September 2026 · Reviewed on 10 October 2026

Oil shock moves the safe haven from gold to the dollar

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — The tightening regime persists: August PCE comes in above 3.7% year on year, Brent stays above $100 and the probability of an October hike holds around 70%. Yields stay high, the dollar strong and gold below its 5 August low.

    What happened

    The triggers did not arrive; the outcomes did. PCE came in at 3.4% on 30 September, below the 3.7% expected, and the probability of an October hike fell to 35%, then to 20% on 2 October. Even so, the 10-year yield peaked at 5.31% on 5 October, the dollar index rose to 102.40 on 8 October and gold fell to $4,111.05 on 7 October, below the $4,146 low. We got the outcome right and the mechanism wrong.

    Evidence: 30 September 2026 · Softer US inflation erodes the odds of an October rate rise
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Data and diplomacy bring relief: PCE comes in below expectations and the probability of an October hike falls below 50%. Rate expectations retreat, the dollar weakens, gold recovers and bond yields fall.

    What happened

    The trigger arrived in full; the outcome never came. PCE came in below expectations at 3.4%, and the October probability fell from 51% to 35%. Yet the 10-year yield closed at 5.29% on 30 September and rose to 5.31% on 5 October. The dollar index reached 102.40 on 8 October, the euro fell from 1.1355 to 1.1186, and gold dropped from $4,157.34 on 30 September to $4,111.05 on 7 October. This is the report's clearest error.

    Evidence: 5 October 2026 · Nasdaq hits record as Treasury yields touch 24-year high
  • Scenario

    Falsified

    The published claim

    H3 · 20% — Growth fears take over: the 2 October jobs report shows sharp weakness or the equity sell-off deepens, and the VIX rises markedly. Yields at the far end of the curve fall, and gold and the dollar both draw safe-haven demand.

    What happened

    The jobs trigger arrived: the US added only 29,000 jobs in September against expectations of about 90,000. Yields at the far end did not fall: FRED data show the 10-year closed at 5.28% on 2 October and hit a 24-year high of 5.31% on 5 October. The VIX stayed low at 15.31 on 2 October, the Nasdaq set a record on 5 October and gold did not recover.

    Evidence: 2 October 2026 · US economy adds only 29,000 jobs in September
  • Watched indicator

    Partly borne out

    The published claim

    Threshold: gold holding below $4,146, its 5 August low, would confirm that the rates channel is suppressing gold's geopolitical premium.

    What happened

    The threshold was breached, but it held for only a week. Investing.com closing prices put gold below it at $4,142.96 on 2 October, $4,139.89 on 5 October, $4,111.05 on 7 October and $4,144.73 on 8 October; it rose above it to $4,164.35 on 6 October. On 9 October it returned to $4,194.24, about $48 above the threshold.

    Evidence: Investing.com — XAU/USD Historical Data
  • Watched indicator

    Confirmed

    The published claim

    Thresholds: after PCE, the US 10-year yield breaking 5.30%, the dollar index breaking 102 and Japan's 30-year yield breaking its September high of 4.22%.

    What happened

    All three thresholds were crossed, two of them only briefly. The 10-year yield crossed at 5.31% on 5 October and fell back to 5.22% on 8 October. Japan's 30-year stood at 4.234% on 5 October and 4.23% on 6 October, then eased to 4.191% on 8 October. The dollar index passed 102 at 102.40 on 8 October. The thresholds pointed the right way but did not settle.

    Evidence: 8 October 2026 · Rupee slides to 96.88 per dollar despite rate hike
  • Causal chain

    Confirmed

    The published claim

    A strong dollar and falling gold erode the dollar value of CBRT reserves; once valuation support fades, the drain on the foreign-currency leg shows up in total and net reserves.

    What happened

    The first two links worked. In CBRT data for the week of 2 October, gold reserves fell by $2.578 billion to $107.062 billion and gross FX reserves fell by $1.237 billion to $60.326 billion. Net reserves excluding swaps dropped from $39.9 billion to $37.9 billion. The third link, the MPC decision and the cost of domestic borrowing, remains open until the October meeting.

    Evidence: 8 October 2026 · Official data put CBRT net reserves excluding swaps at 37.9 billion dollars
  • Falsifier

    Falsified

    The published claim

    Condition that breaks the chain: if a date for Hormuz negotiations is announced and Brent retreats, or if the 30 September PCE print comes in below expectations, the probability of a Fed hike falls, the dollar weakens and the chain stops at the first step.

    What happened

    The condition was met, but the outcome we forecast did not follow. PCE came in below expectations and the October probability fell to 35%. The dollar did not weaken: EUR/USD fell from 1.1355 on 30 September to 1.1186 on 8 October. Nor did the chain stop: CBRT net reserves excluding swaps fell to $37.9 billion. We tied the falsifier to the Fed's October decision; the dollar tracked the real yield, not the October decision.

    Evidence: 30 September 2026 · Softer US inflation erodes the odds of an October rate rise

What we learned

We read the direction of the dollar and gold correctly for a week but got the cause wrong; gold climbed back above the threshold on 9 October. We tied the regime to the probability of a Fed hike in October; when that probability fell from 70% to 20%, the dollar still strengthened and gold still fell. The triggers for both H2 and the falsifier arrived, and in both cases the market moved the other way. The reason is in the FRED data: the entire rise in the 10-year yield came from the real yield, while breakeven inflation held at 2.35%. The regime, in other words, was driven by the real yield and the term premium, not by near-term policy expectations. Change of method: in macro reports, falsifiers will be tied to the real yield and the breakeven rate, not to a single data release or meeting probability.

Published on 2 October 2026 · Reviewed on 9 October 2026

One oil shock meets four different buffers across South Asia

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — The buffers bend but hold: Brent stays in a 95–105 dollar band, Pakistan agrees targeted support through BISP with the IMF, and the RBI moves by no more than 25 basis points on 7 October. The rupee settles around 96, India's reserves hold near 765.9 billion dollars and Türkiye's CDS stays below 250 basis points.

    What happened

    Two of the three triggers held. Brent stayed between 100.20 and 104.13 dollars from 2 to 8 October; the RBI raised the repo rate by 25 basis points to 5.50% on 7 October, and the rupee traded between 96.36 and 96.77. The Pakistan leg is incomplete: talks moved closer to the tranche on 6 October, and targeted support via BISP is planned to start in January 2027, but we have no record of a staff-level agreement. The reserves leg failed, with India's reserves falling 18.34 billion dollars in the week of 25 September to 747.56 billion. Türkiye's CDS was also above 250, at 252–256 basis points on 1–5 October.

    Evidence: 7 October 2026 · RBI raises rates for the first time in nearly four years
  • Scenario

    Still open

    The published claim

    H2 · 30% — Oil above 105 dollars strains the IMF line: Brent settles above 105 dollars, Pakistan fails to resolve its gas arrears, and Sri Lanka extends support. IMF tranches slip, the RBI embarks on two hikes and India's reserves keep falling; in Türkiye the CDS breaks 250 basis points and the dollar/lira passes 50.

    What happened

    The main trigger has not arrived: Brent closed at 104.13 dollars on 8 October, just under 105, with no sign of settling above it. Pakistan's gas arrears deadlock began on 2 October and turned towards a resolution on 6 October. The RBI hiked once, and Malhotra said the next move would be either a hike or a pause. The Türkiye leg is mixed: CDS exceeded 250 on 1–5 October and slipped back to 247.19 on 6 October, while the dollar/lira stood at 49.18, below 50.

  • Scenario

    Falsified

    The published claim

    H3 · 20% — Prices ease and subsidy pressure fades: Gulf exports recover and Brent falls below 95 dollars; the RBI holds at 5.25%. The rupee and Indian equities rebound, and premiums on South Asian dollar debt narrow.

    What happened

    Neither leg of the trigger held. Rather than holding at 5.25%, the RBI voted unanimously on 7 October to raise the rate to 5.50%, its first increase since February 2023. Gulf flows returned to 90% of pre-war levels according to the EIA, but Brent never reached 95 dollars; its lowest close was 100.20 dollars on 7 October. The Sensex and Nifty opened lower on decision day, and Bangladesh's September inflation rose to 8.34%, so price pressure in the region did not fade.

    Evidence: 7 October 2026 · RBI raises rates for the first time in nearly four years
  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: Brent above 105 dollars would show subsidies pushing against IMF ceilings, while a drop below 95 would show rate pressure easing. Türkiye's CDS settling above 250 basis points and the dollar/lira passing 50 would show the energy bill feeding into risk premium and FX demand.

    What happened

    Only one of the four thresholds, CDS, was crossed, and it did not stick. CDS was 252.10 basis points on 1 October, 255.69 on 2 October and 255.29 on 5 October, before easing to 247.19 on 6 October and 247.80 on 7 October. Brent never left the 95–105 dollar band, though at 104.13 dollars on 8 October it came close to the upper threshold. The dollar/lira stayed below 50 at 49.18 on 6 October.

    Evidence: Investing.com — Turkey 5 Year CDS Historical Data
  • Causal chain

    Partly borne out

    The published claim

    The subsidy debate in two IMF-programme countries shifts the tranche calendar; Pakistan's 1.2 billion dollar tranche remains tied to its gas plan and Sri Lanka's 7th review to a staff-level agreement.

    What happened

    The Pakistan leg of the link worked: talks stalled on 2 October over 75 billion rupees of fuel compensation and a proposal to write off 1.4 trillion rupees of gas arrears. Dawn reported on 6 October that the tranche was within reach and that the IMF had made no new demands; whether the delay pushed the tranche back is not yet clear. Against the 22 LNG cargoes sought for winter, only 10–12 have been committed. We have no new record on Sri Lanka's 7th review.

    Evidence: 6 October 2026 · Pakistan nears a $1.2bn IMF tranche as review talks close
  • Causal chain

    Partly borne out

    The published claim

    The region's largest buffer also turns to tightening: if the RBI hikes with CPI at 4.82% and wholesale inflation at 9.92%, higher policy rates and yields become the norm for energy-importing emerging economies.

    What happened

    The first leg held in full: the RBI raised rates by 25 basis points on 7 October, shifted its stance to calibrated tightening by 4 votes to 2, and lifted its FY2027 inflation forecast from 5% to 5.2%. Reserve losses in the two weeks before the decision had approached 33 billion dollars. The contagion leg is not yet proven: we have not measured where India's 10-year yield stands against the 7% threshold, and we have no record of new rate hikes among other energy importers.

    Evidence: 7 October 2026 · RBI raises rates for the first time in nearly four years
  • Falsifier

    Confirmed

    The published claim

    The condition that breaks the chain: if Brent returns to the RBI's assumed 90–95 dollar band, subsidy pressure and the case for a hike fade together; approval of the 10 billion dollar facility Pakistan has sought from the US would break the first link.

    What happened

    Neither leg of the condition materialised, and the chain moved forward. Brent never fell below 100.20 dollars between 2 and 8 October, staying at least 5 dollars above the RBI's assumed band. We have no record of approval for the 10 billion dollar facility Pakistan sought from the US. The RBI hiked as expected, citing crude's September average of 114–116 dollars a barrel as grounds for the decision.

    Evidence: 7 October 2026 · RBI raises rates for the first time in nearly four years

What we learned

The regional read held, but we got two things wrong. First, we gave H3 20% and made an RBI hold at 5.25% its trigger; the decision was a 25 basis point hike and the scenario was falsified. Second, we wrote two legs into H1 that already contradicted the data on publication day. Türkiye's CDS was 252.10 on 1 October, so 'stays below 250' was not true when we published. On the day we wrote that India's reserves would hold near 765.9 billion dollars, a record weekly fall of 18.34 billion was about to be announced. Lesson: scenario bands will be checked against the latest close and the weekly reserves calendar at the time of publication.

Published on 4 October 2026 · Reviewed on 9 October 2026

Record exports leave reserves, not inflation, as the MPC's binding constraint

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — A wide deficit, a slow reserve drain: September CPI comes in close to expectations, Brent stays in a 100–105 dollar band and FX deposit growth slows to below 1 billion dollars a week. Net reserves excluding swaps stay in a 35–40 billion dollar range and the CBRT holds at 37% on 22 October; CDS stays at 245–260 basis points and the 10-year yield at 32.5–33%.

    What happened

    The market and reserve legs held. According to CBRT data released on 8 October, net reserves excluding swaps fell 2 billion dollars to 37.9 billion in the week to 2 October. CDS stayed between 247.19 and 255.69 basis points on 1–7 October, and the 10-year yield between 32.68% and 32.84% on 5–8 October. Brent traded in a 100.20–104.13 dollar band on 2–8 October. The inflation leg of the trigger failed: September CPI rose 1.84% on the month, well below the 2.20% expected. That undercut the assumption of a hold at 37%; the market now expects a cut on 22 October.

    Evidence: Dünya — CBRT data show reserves fell by roughly 4 billion dollars (8 October 2026)
  • Scenario

    Still open

    The published claim

    H2 · 30% — The reserve threshold comes under strain: the energy bill pushes the October deficit above 6 billion dollars, the fund compensation timetable slips and flight into FX deposits accelerates. Net reserves excluding swaps fall below 35 billion dollars; CDS breaks 260 basis points and the 10-year yield 33%.

    What happened

    None of the triggers has materialised yet. Net reserves excluding swaps, at 37.9 billion dollars, remain above 35 billion; CDS peaked at 255.69 on 2 October, below 260. October trade data are due in early November. A 10-article temporary bill on fund compensation was drafted on 6 October, but there is no payment timetable. Karahan said the same day that money leaving the funds had largely gone back into deposits, which suggests the flight is not accelerating.

  • Scenario

    Still open

    The published claim

    H3 · 20% — External balance eases: Brent falls below 95 dollars, the TMSF payment timetable is announced and foreign bond inflows return. The reserve drain stops, the rate-cut debate returns and CDS falls below 240 basis points.

    What happened

    The triggers did not materialise and prices moved the other way: Brent rose to 104.13 dollars on 8 October. No TMSF payment timetable was announced, and CDS did not fall below 240, with its lowest close at 247.19 on 6 October. Only the rate-cut debate returned, and it was driven not by an easier external balance but by inflation below expectations. The 2-year yield fell to 36.42% on 6 October, below the 37% policy rate.

  • Causal chain

    Partly borne out

    The published claim

    FX needs, plus demand for FX deposits from households fleeing the fund crisis, are met from reserves; net reserves excluding swaps keep falling by around 3 billion dollars a week. Indicator: a drop below 37 billion dollars in the 8 October data.

    What happened

    The drain continued, but more slowly than we expected. According to data the CBRT released on 8 October, net reserves excluding swaps fell 2 billion dollars in the week to 2 October, from 39.9 to 37.9 billion; the 37 billion threshold was not breached. Gross FX reserves fell 1.24 billion dollars to 60.33 billion. The household FX demand component weakened: Karahan said on 6 October that small investors were still selling foreign currency and gold.

    Evidence: Dünya — CBRT data show reserves fell by roughly 4 billion dollars (8 October 2026)
  • Causal chain

    Still open

    The published claim

    As the reserve buffer thins, the CBRT loses room to cut on 22 October; CDS and extended-maturity lira yields rise, and the cost of rolling over external debt for the Treasury and banks goes up.

    What happened

    The verdict stays open because the MPC meets on 22 October. Interim signals are working against this link: inflation fell to 29.73% and Commerzbank expects a 100 basis point cut. The 2-year yield stood at 36.53% on 7 October, below the policy rate. CDS eased from 255.29 on 5 October to 247.80 on 7 October. The Treasury, however, borrowed at a compound 39.87% on a 2-year bond on 5 October; domestic borrowing costs remain high.

  • Falsifier

    Partly borne out

    The published claim

    The condition that breaks the chain: if Brent falls durably below 90 dollars, if TMSF fund and CMB interim payments halt the flight into FX deposits, or if foreign flows similar to July's 5.8 billion dollars of portfolio inflows return, the chain breaks at the second step.

    What happened

    The Brent leg did not materialise; the lowest close on 2–8 October was 100.20 dollars. There is no data showing foreign flows returning. On deposits there is a counter-signal: Karahan said on 6 October that money leaving the funds had largely gone back into deposits and that the lira share of deposits had passed 61%. Even so, net reserves excluding swaps fell 2 billion dollars that same week. The flight slowed, but the reserve drain did not stop; the second link weakened without breaking.

    Evidence: 6 October 2026 · Karahan says money leaving funds has largely returned to deposits

What we learned

The base case's reserve range and market bands held, but we set the trigger wrongly. We assumed September CPI would come in close to expectations; at 1.84% it was 0.36 points below. That weakened the assumption of a hold at 37% and pulled rate-cut expectations forward. The second error was pace: we expected a drain of 3 billion dollars a week, the actual figure was 2 billion and the 37 billion threshold was not breached. Reading the reserve channel on its own missed the rate-expectations channel; we had written the same lesson in the scorecard for the 2 October CBRT report, and it was not applied when this report was written.

Published on 3 October 2026 · Reviewed on 9 October 2026

Russian diesel has left the market, and G7 stocks only buy time

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — The bridge works, the gap persists: the G7 release proceeds on schedule, the US imposes no export ban and Russia extends its ban into November. Diesel retreats from its record but does not return to pre-war levels; Brent swings around 100 dollars and diesel margins stay high.

    What happened

    Two legs held. On 7 October IEA members decided to accelerate the remaining roughly 100 million barrels and bring diesel forward; as of 8 October we have no record of a US diesel export restriction. The EU weighted average diesel price was 2.149 euros a litre on 5 October, still 35% above the 1.594 euros of 23 February. Brent stayed in a 100.20–104.13 dollar band from 2 to 8 October. Whether Russia's ban runs into November will be clear on 31 October.

    Evidence: 7 October 2026 · IEA members speed up emergency oil stock releases and prioritise diesel
  • Scenario

    Still open

    The published claim

    H2 · 30% — The US narrows the export door: US pump prices keep rising and Washington restricts diesel exports. Europe and Latin America turn to distant supply, supply outages begin in poorer importers, and diesel margins and clean tanker freight rise sharply.

    What happened

    As of 8 October we have no record of a formal export restriction from Washington; on 2 October, after the G7 release, Trump said a ban had never been on the table. The pressure continues: US commercial crude stocks fell from 427.3 to 424.1 million barrels in the week to 2 October. The EIA expects East Coast distillate stocks to stay 20–30% below the five-year average through the winter. Storm Isaias shut 25.08% of Gulf of Mexico output on 7 October.

  • Scenario

    Still open

    The published claim

    H3 · 20% — Russian supply returns: Russia lifts its ban on 31 October, or an energy ceasefire halts the refinery strikes. Russian diesel returns to the Mediterranean and Turkish markets; diesel margins narrow and Brent stays below 100 dollars.

    What happened

    The verdict stays open because the deadline is 31 October, but as of 8 October events are running against the scenario. Ukraine said on 4 October it had knocked out 51% of Russian refining capacity; the figure has not been independently verified. According to the IEA, Russian diesel output has fallen by about 30%. On 6 October 650 drones flew at Moscow and a fire broke out at an airport fuel depot. We have no record of an energy ceasefire.

  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: a sustained fall in Brent from 102.25 dollars on 2 October to below 100 would show the stock release reaching prices in the first tranche; weekly declines in US commercial stocks despite the release would show mounting pressure in Washington for an export ban.

    What happened

    The stocks threshold was met: US commercial crude stocks fell 3.2 million barrels to 424.1 million in the week to 2 October. The price threshold was not: Brent came close to 100 dollars at 100.32 on 5 October and 100.20 on 7 October but did not break below it, then jumped to 104.13 dollars on 8 October. The first 20-day tranche of the release runs to 22 October, so the price test is not over.

    Evidence: EIA — Weekly Petroleum Status Report highlights
  • Causal chain

    Partly borne out

    The published claim

    A diesel-heavy release in the first 20 days temporarily lowers European prices, but an average flow of 0.83 million barrels a day cannot fully replace lost Russian diesel, and stocks draw down. Indicator: the EU diesel average falling below the 2.26 euros a litre of 17 September.

    What happened

    The price leg materialised, but the causal link is weak. The EU weighted diesel average fell to 2.149 euros a litre in the 5 October bulletin, below the 2.26 euro marker. Because the G7 decision was taken on 2 October, it is impossible to separate how much of the decline came from the release. The stock-draw leg finds support: the EIA estimates global inventories drew 1.9 million barrels a day in the third quarter and expects a further 0.7 million barrels a day in the fourth.

    Evidence: fuel-prices.eu — 2026 fuel price shock: EU pump prices (EU Weekly Oil Bulletin)
  • Falsifier

    Still open

    The published claim

    The condition that breaks the chain: if Russia lifts its diesel ban on 31 October and repaired refineries return to exports, or Hormuz transits move durably back towards normal, the chain stops at the second step.

    What happened

    The verdict stays open because the deadline is 31 October. The Hormuz leg is mixed: transits rose on 4 October despite the attacks, but on 7 October Iran threatened to close routes along the Omani coast and a tanker was hit off Qatar. The refinery leg is moving the other way; Ukrainian strikes continue and, according to the IEA, Russian diesel output has fallen by about 30%. That makes it harder for the condition to be met by 31 October.

What we learned

No scenario has reached its deadline yet, but two set-up errors are already visible. We gave a US export restriction 30%, yet Trump had said on 2 October, a day before publication, that a ban was never on the table. That probability was too high given what was known on publication day. Second, we treated EU diesel falling below 2.26 euros as the marker of the release; the price did fall to 2.149 euros on 5 October, but the decline can also be explained independently of the G7 decision. Lesson: policy scenarios will be tested against statements from the final 48 hours before publication, and price markers will be chosen net of the pre-intervention trend.

Published on 3 October 2026 · Reviewed on 9 October 2026

As buyers at the far end thin out, Britain shortens its borrowing

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 50% — Holding pattern at elevated levels: energy prices move sideways, the 28 October budget keeps the fiscal rules intact and the DMO holds the long share near 9.1%. The 30-year yield stays in a 5.8–6.1% band and the BoE holds on 5 November on a split vote.

    What happened

    The verdict stays open because the budget falls on 28 October and the BoE decision on 5 November. The yield leg is within its band for now: the 30-year gilt rose 13 basis points to 6.036% on 7 October, its highest since 1998. The Chancellor reaffirmed the commitment to the fiscal rules on 6 October. The energy leg is under strain: Brent jumped to 104.13 dollars on 8 October. The market is pricing roughly 30 basis points of tightening by year-end.

  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — The buyer constraint deepens: the budget fails to close the fiscal gap convincingly or energy prices rise again, and the US 30-year yield breaks 5.65%. The 30-year gilt closes above 6%, the DMO cuts long issuance further and the BoE hikes in November.

    What happened

    The market leg of the trigger arrived: the US 30-year yield rose to 5.66% on 5 October and on 7 October hit its highest level since 2002. Gilts followed; the 30-year yield touched 6.036% intraday on 7 October, beating the 1 October record, but we could not confirm a close above 6%. Bank of America expects the borrowing requirement to rise by 15 billion pounds. The DMO remit comes on 28 October and the BoE decision on 5 November; those legs remain open.

    Evidence: 7 October 2026 · Britain's long-term borrowing costs hit their highest since 1998
  • Scenario

    Still open

    The published claim

    H3 · 20% — Energy eases, the far end relaxes: Brent futures retreat, October inflation comes in below expectations and the 28 October budget lowers the borrowing requirement. The 30-year yield falls below 5.6% and expectations of a BoE hike are pushed back.

    What happened

    The deadline has not passed, but every leg moved the wrong way in the five trading days after publication. Brent rose from 102.25 dollars on 2 October to 104.13 dollars on 8 October. Rather than approaching 5.6%, the 30-year yield set a record of 6.036% on 7 October. The BoE's Mann said on 6 October that inflation had become embedded in the economy; the market is pricing more than 100 basis points of hikes by the end of 2027.

  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: the US 30-year yield breaking 5.65% would show the global sell-off at the far end resuming; the German 30-year above 4% would show the buyer constraint spreading to Europe's core; the Japanese 30-year settling above 4.3% would show another source of global extended-maturity demand withdrawing.

    What happened

    Only the US threshold of the three was crossed. The US 30-year yield reached 5.66% on 5 October and hit its highest since 2002 on 7 October. The German 30-year stayed in a 3.852–3.866% band on 5–8 October and came nowhere near 4%. The Japanese 30-year peaked at 4.234% on 5 October and eased to 4.191% on 8 October; 4.3% was not breached. For now the selling is concentrated at the Anglo-Saxon far end.

    Evidence: 7 October 2026 · Britain's long-term borrowing costs hit their highest since 1998
  • Causal chain

    Still open

    The published claim

    Every BoE hike against energy-driven inflation swells the Treasury's interest bill faster; tension between monetary and fiscal policy rises and the risk premium at the far end becomes entrenched.

    What happened

    The verdict stays open because the BoE decides on 5 November. Interim signals support the link: Mann said on 6 October that inflation could reach 4% around the anniversary, and she voted for a hike in July and September. The 10-year gilt yield rose to 5.48% on 7 October, its highest since 2007. Every rise in yields enlarges the savings the budget needs to meet the fiscal rules.

  • Falsifier

    Still open

    The published claim

    The condition that breaks the chain: if energy prices fall and inflation drops below BoE projections, the pressure to hike lifts. A new buyer base making the 87 billion pounds of demand seen on 8 September durable would also break the chain at the first link.

    What happened

    The verdict stays open because the first link depends on the 28 October DMO remit. As of 8 October neither leg of the condition has materialised. Brent is rising at 104.13 dollars, and Mann says inflation has become embedded. We have no record of a new extended-maturity buyer base in the UK. For comparison, foreign-heavy buyers took 80.3% of the US 10-year auction on 7 October.

What we learned

No scenario has reached its deadline, but the first week showed two things. The core read, that the UK far end is under a buyer constraint, hardened: the 30-year yield set a new record of 6.036% four days after publication. The global contagion thesis, by contrast, stayed weak; the German and Japanese thresholds were not crossed, and the selling concentrated in the US and the UK. The 20% we gave H3 weakened with every data point after publication. Lesson: far-end scenarios need country-specific constraints and the global term premium tracked through separate indicators.

Published on 1 October 2026 · Reviewed on 8 October 2026

The energy shock cools in America and heats up in Europe

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Divergence becomes entrenched: euro-area September data also come in higher, US data are moderate and the Fed holds on 28 October. The ECB keeps a hike on the table while the Fed waits. EUR/USD heads above 1.1350 and the French spread stays at 100–120 basis points.

    What happened

    The data legs arrived: euro-area September inflation hit a three-year high of 3.8%, while the US added only 29,000 jobs in September. The Fed decides on 28 October, so that leg is still open. The market legs went the other way. EUR/USD fell to 1.1232 on 1 October and 1.1259 on 2 October, never reaching 1.1350. The French spread broke above the band, at 130 basis points on 1 October and about 129 on 6 October.

    Evidence: 2 October 2026 · Energy drives euro area inflation to a three-year high
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Energy peaks: December Brent stays around or below $98.03 and the energy component slows in October flash data. The ECB and the Fed both wait. Long-dated yields give back part of September's jump.

    What happened

    The trigger did not arrive: December Brent rose to $102.31 on 1 October and did not fall below $100 through 7 October. The market leg also went the wrong way, with the US ten-year yield reaching a 24-year high of 5.31% on 5 October. The effect is coming for a different reason. On 6 October the ECB's Nagel and Lane said they saw no rush to hike in October. Their rationale was not an energy peak but the absence of second-round effects.

    Evidence: 6 October 2026 · ECB's Nagel and Lane see no rush to hike in October
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — The French crisis spreads: a confidence vote or the fall of the government comes onto the agenda before the 17 November vote, and the spread breaks durably above 120 basis points. Fiscal stress jumps to Italian and Spanish bonds and the euro weakens.

    What happened

    The spread leg materialised a day after publication: it broke above 130 basis points on 1 October and stood at about 129 on 6 October, with France's ten-year at 4.73% and Germany's at 3.45%. The euro also weakened. The political trigger did not arrive: as of 7 October we have no record of a confidence vote or a government collapse, and Le Pen presented a counter-budget on 6 October. We have no data measuring spillover to Italy and Spain.

    Evidence: 6 October 2026 · Le Pen's counter-budget promises €140bn in net savings
  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: the US ten-year yield settling above 5.29% would show the term premium still rising; EUR/USD above 1.1350 would show ECB–Fed divergence being priced; December Brent above $103.50 would show energy pressure persisting.

    What happened

    Of the three thresholds, only the yield threshold was crossed, and it did not hold. The ten-year yield was 5.31% on 5 October and 5.27% on 6 October. EUR/USD stood just below the threshold at 1.1345 on 30 September, then slipped to the 1.12s. December Brent peaked at $102.31 between 1 and 7 October, staying below $103.50.

    Evidence: FRED — 10-Year US Treasury yield (DGS10)
  • Causal chain

    Partly borne out

    The published claim

    The ECB keeps the option of a further hike open; euro-area front-end yields rise and pressure builds on a French spread already at 111–120 basis points.

    What happened

    The spread link worked: the French spread rose from 111.2 basis points on 30 September to above 130 on 1 October. The ECB link was only half complete. Nagel and Lane played down an October hike, but markets still put the probability of a December hike at about 66%. We have no series measuring euro-area front-end yields.

    Evidence: 6 October 2026 · ECB's Nagel and Lane see no rush to hike in October
  • Falsifier

    Partly borne out

    The published claim

    The condition that breaks the chain: if December Brent stays around $98.03 and the energy component slows in October data, the ECB waits and the chain breaks at the first link. A French spread below 100 basis points would weaken the second link.

    What happened

    The condition was not met: December Brent traded in a $100.32–102.31 range and the spread stood at 129 basis points. Yet the outcome of the condition arrived on its own. On 6 October the ECB signalled room to wait in October. The first link weakened without the breaking condition we had set. This shows the falsifier was tied to the wrong variable.

    Evidence: 6 October 2026 · ECB's Nagel and Lane see no rush to hike in October

What we learned

We read the divergence from central banks' tone, but the currency tracked US yields. While the ECB looked more hawkish, the euro fell from 1.1345 to the 1.12s because the US ten-year hit a 24-year high. We tied the ECB's pause to an energy peak; the rationale turned out to be the absence of second-round effects. On the morning we wrote the 100–120 band for France, the reading was already at 120, and a day later it was 130. Lesson: currency scenarios will be tied to the yield differential itself, not to policy tone. Any threshold or band will be set at least one day's volatility away from the last reading on the morning of publication.

Published on 1 October 2026 · Reviewed on 8 October 2026

As Hormuz heats up, oil's real risk shifts to the Red Sea

Open the report ›
  • Scenario

    Falsified

    The published claim

    H1 · 55% — Concentration on a single strait persists: the East-West line stays in a 2.0–2.65 million barrels a day range, Bab el-Mandeb flows hover around 2.56 million barrels and attacks in Hormuz are sporadic but persistent. Backwardation holds and freight and war-risk premiums stay high.

    What happened

    The scenario's defining condition broke down within a week. On 6 October the Saudi energy minister said the East-West line was back at 5.8 million barrels a day; according to AA, flows had already exceeded 80% of capacity a week earlier. Hormuz's share of Gulf crude exports excluding Iran fell from 83% before the war to 60%. Attacks were not sporadic but daily from 2 October. Only the market leg held: VLCC rates reached $1.2m a day.

    Evidence: 6 October 2026 · Saudi East-West pipeline returns to 5.8 million barrels despite attack claim
  • Scenario

    Partly borne out

    The published claim

    H2 · 25% — The Red Sea outlet opens: the East-West line nears full capacity, Yanbu loadings continue without attacks and daily transits through Bab el-Mandeb rise well above 26 ships. The prompt premium erodes and the December contract could fall towards $95.

    What happened

    The physical trigger largely arrived. The line stood at 5.8 million barrels on 6 October, against a maximum capacity of 7 million. Windward counted 42 transits through Bab el-Mandeb on 4 October, and IMF PortWatch 30 and 34 on 3–4 October. OilPrice says only 60% of Gulf crude exports excluding Iran now pass through Hormuz. The no-attack condition is unclear, as the reported 5 October attack on the Khurais pumping station could not be verified. The price leg did not arrive: December Brent bottomed at $100.32 on 5 October, nowhere near $95.

    Evidence: 5 October 2026 · Estimates of oil flowing through Hormuz diverge by up to fivefold
  • Scenario

    Still open

    The published claim

    H3 · 20% — Both fronts seize up: attacks cut Hormuz transits below 9 ships, while a new hijacking in the Gulf of Aden or a new attack in the Red Sea occurs in the same period. The December contract breaks above $103.50 and backwardation steepens.

    What happened

    We set no deadline for this tail scenario, so the verdict stays open. As of 7 October the two outlets had not narrowed at the same time. Windward counted 13 Hormuz transits on 4 October. On the Red Sea side, explosions were reported near a tanker 60 nautical miles south of Mokha on 4 October, but the ship was undamaged. December Brent stayed in a $100.32–102.31 range between 1 and 7 October and did not reach $103.50.

  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: daily Bab el-Mandeb transits above 26 ships would show the Red Sea regaining its role as a back-up route; fewer than 9 in Hormuz would signal a blockage; December Brent above $103.50 would mean lasting scarcity, and below $95 that the premium has unwound.

    What happened

    The Bab el-Mandeb threshold was crossed and showed what we said it would: Windward counted 23 entries and 19 exits on 4 October. The Hormuz threshold was poorly set: Windward gave 13 transits on 4 October while PortWatch showed 4 the same day. PortWatch has been below 9 every day since mid-September. Neither Brent threshold was crossed; the December contract stood at $100.20 on 7 October.

    Evidence: 4 October 2026 · Hormuz tanker attacks turn almost daily, yet transits still rise
  • Causal chain

    Partly borne out

    The published claim

    Because the back-up outlet has weakened, every attack in Hormuz threatens larger volumes and feeds into freight; because the Red Sea has narrowed, scarcity is felt in diesel before crude; Türkiye's energy bill stays high even if headline Brent falls.

    What happened

    The premise of the first link reversed: cargo shifted back from Hormuz to the Red Sea. Even so, the cost link worked. VLCC rates are about $1.2m a day, against $80,000 a year ago. Aramco widened its Asian discount for November from $2 to $5. The product leg also held: Gulf refined-product exports in September were only 60% of pre-war levels. There are no data yet for the Türkiye bill link.

    Evidence: 5 October 2026 · Aramco cuts November Asia price to deepest discount in six years
  • Falsifier

    Falsified

    The published claim

    The condition that breaks the chain: if the East-West line nears its 7 million-barrel capacity and daily Bab el-Mandeb transits rise well above 26 ships, cargo spreads across two outlets and the chain breaks at the first step.

    What happened

    Both legs of the breaking condition were met on 4–6 October. The East-West line returned to 5.8 million barrels on 6 October, about 83% of capacity. On 4 October Windward counted 42 transits through Bab el-Mandeb and PortWatch 34. The core of our reading, that risk was shifting to the Red Sea, lost its validity six days after publication.

    Evidence: 6 October 2026 · Saudi East-West pipeline returns to 5.8 million barrels despite attack claim

What we learned

We built the main scenario (55%) on a temporary outage, and we were wrong. The East-West line had reopened on 22 September; the 2.0–2.65 million-barrel range was a snapshot of the repair process, not a new equilibrium. The line was back at 5.8 million barrels within six days. The second error lay in the thresholds: we took the 26-ship threshold for Bab el-Mandeb and the 9-ship threshold for Hormuz from a different count. In our own PortWatch series, Bab el-Mandeb was already at 27–37 in the week before publication. Lesson: the throughput of infrastructure under repair will not be used as a scenario baseline. Every threshold will be tied to a single, named series and checked against that series' latest value on the day of publication.

Published on 2 October 2026 · Reviewed on 8 October 2026

Edge devices become a shared doorway as patch windows shrink to three days

Open the report ›
  • Scenario

    Still open

    The published claim

    H1 · 55% — High tempo, limited pricing: 15–20 edge-product entries a month continue to join KEV in October, but no incident causes a service outage in critical infrastructure. The VIX stays below 20 and the decline in cyber insurance pricing narrows to less than 4%.

    What happened

    The deadline is the end of October, so the verdict stays open. The CISA catalogue was last updated on 4 October and holds four October entries. Two of them are edge products: FortiMail on 1 October and NetScaler on 4 October. That pace is far below 15–20 edge entries a month. In September 18 of 43 entries were edge products, so the 15–20 band amounted to treating a record month as the floor. The VIX was below 20 at 15.01 on 6 October.

    Evidence: CISA — Known Exploited Vulnerabilities catalogue (JSON)
  • Scenario

    Still open

    The published claim

    H2 · 30% — An edge incident turns into systemic cost: a widespread VPN or SD-WAN flaw causes simultaneous outages at several organisations in energy, transport or finance. The VIX breaks above 20 and cyber insurance pricing ends 12 quarters of decline.

    What happened

    As of 7 October the trigger had not arrived. The biggest outage was in Osaka, where the virtualisation layer went down on 2 October and about 500 servers and their backups were encrypted. But the target was a university and the entry point was not disclosed. In Korea data were stolen from seven financial institutions without any service outage. The VIX traded in a 15.01–16.39 range between 1 and 6 October.

    Evidence: 6 October 2026 · Ransomware in Osaka locks 500 servers and the backups with them
  • Scenario

    Still open

    The published claim

    H3 · 15% — Tempo falls: September's 43 entries prove a one-off, and KEV's monthly volume in October and November drops below the 2026 average of about 25. Pressure on the patch window eases.

    What happened

    The deadline is the end of November, so the verdict stays open. The first week supports the scenario to which we gave the lowest probability. The catalogue has four entries between 1 and 4 October and no update after 4 October. The monthly average for January–August was 25.4, against 43 in September. If this pace holds, October will close below 25; we are not ruling on a week of data.

    Evidence: CISA — Known Exploited Vulnerabilities catalogue (JSON)
  • Watched indicator

    Still open

    The published claim

    Threshold: the VIX, at 16.39 on 1 October, breaking above 20 would show an edge-device or ransomware incident moving beyond a single company's losses into a market-wide risk premium.

    What happened

    The threshold was not crossed. FRED data show the VIX at 15.31 on 2 October, 15.52 on 5 October and 15.01 on 6 October. In the same week 500 servers were locked in Osaka and breaches were confirmed at seven institutions in Korea; neither registered at the market level. No incident large enough to test whether the threshold works has yet occurred.

    Evidence: FRED — VIX (VIXCLS)
  • Causal chain

    Partly borne out

    The published claim

    Attackers entering through unpatched edge devices cause outages in the commercial IT layer and the cost falls directly on the business; the number of edge entries added to KEV and breach disclosures are the first indicators.

    What happened

    The outage link worked, but our description of the entry point was too narrow. On 4 October a new zero-day in NetScaler hit devices that had already been patched; it is the fourth NetScaler entry in the catalogue since September. Patched devices, in other words, are also a way in. In the other incidents the entry point was not an edge device. The attackers used Zammad help-desk software and, in Korea, external websites and sales-support servers.

    Evidence: 4 October 2026 · Patched NetScaler devices hit again by a new zero-day
  • Falsifier

    Still open

    The published claim

    The condition that breaks the chain: if KEV's monthly volume falls below about 25 in October and November and no major service outage occurs, the second link, a turn in cyber insurance pricing, does not form.

    What happened

    The deadline is the end of November, and the Marsh third-quarter index that will measure the insurance link has not yet been published; cyber pricing fell 4% in the second quarter. The first leg of the condition is moving towards fulfilment in the first week, with four entries in October up to 4 October. As of 7 October we have no record of a major service outage in critical infrastructure.

What we learned

There are not yet enough data for a verdict after one week, but two weaknesses are already visible. We wrote 15–20 edge entries a month into the main scenario; that band meant treating September's 18, the only record month of 2026, as the floor. The first four days of October brought two edge entries. Second, we narrowed the entry point to edge devices. In October's incidents the way in was help-desk software, external websites and patched NetScaler devices. Lesson: tempo scenarios will be tied to an average of at least six months, not to a single record month. The chain will also cover cases where patching itself is not enough.

Published on 1 October 2026 · Reviewed on 7 October 2026

The energy shock splits the two sides of the Atlantic

Open the brief ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Divergent shock: euro-area data beat expectations while US hiring is moderate; the Fed holds on 28 October and the ECB stays tight. December Brent holds in a $95–102 band, the US long end stays at 5.2–5.4% and the euro remains under pressure.

    What happened

    The European leg held. Euro-area September inflation reached 3.8% on 2 October, the highest since 2023, and euro/dollar fell from 1.1355 on 30 September to 1.1204 on 5 October. The US 10-year yield stayed inside the band at 5.31% on 5 October. US hiring came in weak rather than moderate: just 29,000 jobs in September. At the ECB, Nagel and Lane said on 6 October they saw no rush to hike in October, softening the 'stays tight' leg. The Brent band is borderline: the December contract breached the upper bound at $102.31 on 1 October, then closed at $100.32 and $101.08. The Fed decides on 28 October; that leg is still open.

    Evidence: 2 October 2026 · Energy drives euro area inflation to a three-year high
  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — Second oil wave: attacks intensify in Hormuz or the Red Sea and the December contract tops $100; odds of an October hike climb back above 50%, long-end yields test new highs and Türkiye CDS settles above 250 basis points.

    What happened

    The trigger arrived. Three tankers were hit in Hormuz on 3–4 October and attacks became almost daily. December Brent has closed above $100 every session since 1 October: $102.31, $102.25, $100.32, $101.08. Türkiye CDS closed at 252.10 basis points on 1 October, 255.69 on 2 October and 255.29 on 5 October. The US 10-year yield touched 5.35% intraday on 5 October. The rates leg worked in reverse: after the 2 October payrolls, the odds of an October hike fell to 20% instead of rising above 50%.

    Evidence: 4 October 2026 · Hormuz tanker attacks turn almost daily, yet transits still rise
  • Scenario

    Falsified

    The published claim

    H3 · 20% — Easing window: Iran–US talks via Qatar make concrete progress, the East-West line nears capacity and December Brent drops below $95; the oil curve flattens and Türkiye CDS eases.

    What happened

    The trigger did not arrive. No progress has been published on the Qatar channel, and two more tankers were hit in Hormuz on 4 October. December Brent never fell to $95; its lowest close was $100.32 on 5 October. Türkiye CDS did not ease and stayed above 250 basis points.

    Evidence: Investing.com — Brent futures historical data
  • Falsifier

    Still open

    The published claim

    The confluence reading (the bill for the energy shock is passing from state budgets to households) is refuted if at least two of Türkiye, Italy, Pakistan and India announce a new fuel tax cut, price cap or broad subsidy by the end of October.

    What happened

    Due 31 October; verdict open. As of 7 October we have no record of any of the four announcing a new fuel tax cut or broad subsidy. One sign tests the edge of the reading: Sri Lanka restored its diesel subsidy on 2 October despite an IMF warning, but Sri Lanka is not among the countries named in the falsifier. In Italy, Meloni sought budget flexibility rather than fuel support.

  • Watched indicator

    Confirmed

    The published claim

    Primary watch: euro-area flash inflation on 1 October. A beat would entrench ECB–Fed divergence; a France–Germany spread above 120 basis points and the euro below 1.1350 would show the pressure is no longer country-specific.

    What happened

    All three conditions were met. Euro-area September inflation came in at 3.8%. On 1 October the French 10-year yield closed at 4.924% and the German at 3.520%, a spread of about 140 basis points; at 127 on 6 October it was still above 120. Euro/dollar fell to 1.1298 on 1 October and 1.1204 on 5 October.

    Evidence: Investing.com — France 10-year bond yield historical data
  • Watched indicator

    Partly borne out

    The published claim

    Secondary watch: CBRT reserve data on 1 October. Gross reserves falling below the $171bn in our calculation would confirm that the fund crisis and the energy bill are jointly driving demand for foreign currency.

    What happened

    In the official data of 1 October, gross reserves stood at $171.2bn for the week of 25 September; the threshold was not crossed that day. Net reserves excluding swaps, however, shed $3.2bn and fell below $40bn. The following week, market estimates put gross reserves at $167.2bn, extending the losses to a sixth week; official data are due on 8 October.

    Evidence: 5 October 2026 · CBRT gross reserves fall for a sixth straight week

What we learned

The European call was right: inflation came in at 3.8%, the French spread rose above 120 basis points and the euro fell below 1.1350. Where we went wrong was the probability split. The trigger for H2, to which we gave 30%, arrived within a week: attacks intensified, Brent stayed above $100 and Türkiye CDS passed 250. But H2's assumption that the Fed would be pushed towards a hike worked in reverse; with payrolls at 29,000, the odds fell to 20%. After the 29 and 30 September briefs, this was the third time we tied an oil shock to a Fed hike, and all three times the data said the opposite. From now on the oil and rates legs will not share a scenario; the US rate path will be read from a separate trigger, based on payrolls and price indices.

Published on 2 October 2026 · Reviewed on 7 October 2026

Factory price pressure jumps, yet bond markets still pull back

Open the brief ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Sticky costs, piecemeal easing: Brent holds in a $95–110 band, the ISM price index stays above 70 and the Fed holds on 28 October. US 10-year at 5.0–5.4%, French spread at 120–150 basis points, a managed depreciation of dollar/lira.

    What happened

    Every measurable leg held. December Brent closed at $100.32 on 5 October and $101.08 on 6 October. The ISM services price index hit its highest since 2022 on 5 October and stayed above 70. The US 10-year yield was 5.31% on 5 October. The France–Germany spread was about 137 basis points on 5 October and 127 on 6 October. Dollar/lira stood at 49.18 on 6 October. The Fed leg is open: futures put the odds of no change in October at about 78%, with the decision on 28 October.

    Evidence: 5 October 2026 · US services prices climb at fastest pace since 2022
  • Scenario

    Falsified

    The published claim

    H2 · 25% — Hormuz eases and easing is vindicated: a temporary Iran–US arrangement emerges via Qatar, attacks in Hormuz stop and Brent falls below $95; emerging-market CDS eases.

    What happened

    The trigger did not arrive. Attacks intensified instead of stopping: one tanker was hit on 3 October and two on 4 October. December Brent did not fall below $95, closing at $101.08 on 6 October. Türkiye CDS did not ease, standing at 255.29 basis points on 5 October.

    Evidence: 4 October 2026 · Hormuz tanker attacks turn almost daily, yet transits still rise
  • Scenario

    Falsified

    The published claim

    H3 · 20% — Easing backfires: Brent tops $110, net reserves excluding swaps fall below $35bn and the France–Germany spread passes 150 basis points; dollar/lira breaks 50 and Türkiye CDS climbs above 250.

    What happened

    The two measurable legs of the trigger did not arrive. December Brent's highest close was $102.25. The France–Germany spread peaked at about 140 basis points on 2 October and eased to 127 on 6 October, never reaching 150. Dollar/lira held at 49.18. Türkiye CDS is above 250, but it was already 252.10 on the day of publication. Official data on net reserves excluding swaps are due on 8 October.

    Evidence: Investing.com — France 10-year bond yield historical data
  • Falsifier

    Still open

    The published claim

    The confluence reading (easing is quietly spreading in response to the energy shock) is refuted if Italy gets no extra flexibility at the 8–9 October Eurogroup, the IMF secures a subsidy-free deal in Pakistan and the CBRT tightens funding again at the 22 October MPC.

    What happened

    Not yet due. The Eurogroup meets on 8–9 October and the CBRT decides on 22 October. The IMF review in Pakistan was expected to wrap up by 7 October; as of 6 October nine tables were still open, and the government plans to move targeted energy support to BISP in January 2027. On the CBRT side, September inflation of 1.84% has put a cut rather than tightening on the table, which supports the reading.

  • Watched indicator

    Confirmed

    The published claim

    Primary watch: US September payrolls on 2 October. Weak data would strengthen the growth-loss reading and push October hike odds below 30%; strong data would bring ISM's 77.9 price index to the fore.

    What happened

    The data came in weak: 29,000 jobs in September, with the previous two months revised down by a combined 60,000. October hike odds fell from 64% a week earlier to 20%, below the 30% threshold we had set.

    Evidence: 2 October 2026 · US economy adds only 29,000 jobs in September
  • Watched indicator

    Confirmed

    The published claim

    Secondary watch: TurkStat September inflation on 3–5 October. A print above the expected 2.20% month on month would call the CBRT's easing of SME credit into question before the 22 October MPC.

    What happened

    TurkStat reported 1.84% month on month and 29.73% year on year on 5 October; expectations were not exceeded. Rather than questioning the easing, markets read it as room to extend it: Commerzbank expects a 100 basis-point cut on 22 October. The core C index, however, rose 2.14% month on month, above the headline.

    Evidence: 5 October 2026 · Softer September inflation puts a rate cut on the table

What we learned

This brief was the week's best-calibrated text: H1's measurable legs held and the payrolls threshold was well set. But we wrote loosely in two places. For H3's market impact we wrote 'Türkiye CDS climbs above 250'; CDS was already at 252.10 on the day of publication, so the item tested nothing. H1's Fed leg depends on 28 October and cannot be closed in a one-week check. Lesson: scenario thresholds will be set against the last close at the time of publication, and every scenario will carry an explicit expiry.

Published on 2 October 2026 · Reviewed on 7 October 2026

The CBRT eases with one hand while reserves drain from the other

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Targeted easing, currency held with reserves: September CPI comes in around 2.20%, weekly losses in net reserves excluding swaps stay under $3bn and the CBRT holds at 37% on 22 October. Dollar/lira 49–50, CDS below 250, 10-year yield below 33%.

    What happened

    The currency and bond legs held: dollar/lira was 49.18 on 6 October, and the 10-year yield 32.68% on 5 October and 32.82% on 6 October. The CDS leg failed: 255.69 basis points on 2 October and 255.29 on 5 October; it never fell below 250. CPI did not land around 2.20% but well below, at 1.84% month on month. That also weakened the assumption of a hold at 37%; a cut is being discussed for 22 October.

    Evidence: Investing.com — Türkiye 5-year CDS historical data
  • Scenario

    Still open

    The published claim

    H2 · 35% — Easing leaks into dollars: retail FX deposits keep rising by more than $1bn a week, net reserves excluding swaps fall below $35bn and the CBRT reverses easing on 22 October. Dollar/lira tops 50, CDS settles above 250.

    What happened

    Data for two legs of the trigger are not yet out: FX deposits and net reserves excluding swaps for the week of 2 October are due on 8 October. Market estimates show gross reserves falling from $171.2bn to $167.2bn that week, extending the losses to a sixth week. The market legs are mixed: CDS has settled at 255 basis points, while dollar/lira stayed below 50 at 49.18.

  • Scenario

    Partly borne out

    The published claim

    H3 · 15% — Confidence is repaired: a payment timetable is announced for the 66 funds outside the scheme, September CPI undershoots and foreign bond outflows stop. The 10-year yield falls below 32% and CDS to around 240 basis points.

    What happened

    The inflation leg of the trigger arrived: TurkStat reported 1.84% month on month on 5 October against expectations of 2.20%. Markets did not respond: the 10-year yield stayed in a 32.68–32.82% range and CDS was 255.29 basis points on 5 October. We have no record of a payment timetable for the 66 funds; the number of detainees in the investigation rose to 85 on 3 October.

    Evidence: 5 October 2026 · Softer September inflation puts a rate cut on the table
  • Watched indicator

    Partly borne out

    The published claim

    Thresholds: dollar/lira above 50 would mark the end of currency calm; CDS settling above 250 basis points would show reserve erosion feeding into the risk premium; a 10-year yield staying above 33% would show bond outflows continuing.

    What happened

    Of the three thresholds, only CDS was breached: 252.10 basis points on 1 October, 255.69 on 2 October and 255.29 on 5 October. Dollar/lira stayed below 50 at 49.18 on 6 October. The 10-year yield stayed below 33%, at 32.68% on 5 October and 32.82% on 6 October.

    Evidence: Investing.com — Türkiye 5-year CDS historical data
  • Causal chain

    Still open

    The published claim

    Interim payouts and a looser reserve-requirement ratio release lira liquidity, part of it moves into dollars and FX deposits rise by more than $1bn a week; demand is met from reserves, and if net reserves excluding swaps fall below $35bn the CBRT reverses easing on 22 October.

    What happened

    Official data to test the links in the chain are due on 8 October. The only sign we have supports the direction of the final link: market estimates show gross reserves down another $4bn in the week of 2 October. Commerzbank says the lira is being supported by central bank and state bank intervention. There are no new figures on FX deposits or interim payout flows.

  • Falsifier

    Still open

    The published claim

    Condition that breaks the chain: if September CPI undershoots and foreign bond outflows stop, the return on holding lira offsets demand for dollars and the second link breaks.

    What happened

    The first half of the condition was met: September CPI came in at 1.84% month on month, below the 2.20% expected. Data on the second half, foreign bond flows for the week of 2 October, are due on 8 October. Because lower inflation has brought forward rate-cut expectations, the return on lira may fall rather than rise. That undermines the logic on which we built the breaking condition.

What we learned

We wrote the risk-premium leg of the base case (50%) with a threshold already contradicted on publication day: we said CDS would stay below 250, yet it closed at 252.10 on 1 October and has not been below 250 since. There is a second error. We treated an inflation undershoot as a condition that repairs confidence. The print came in at 1.84%, but CDS and bond yields did not move. Lower inflation also brought forward rate-cut expectations, which can cut the return on holding lira and strengthen the currency link in the chain. Lesson: scenario thresholds will be set against the last close at the time of publication, and the effect of inflation data on the currency will be read through the rate-expectations channel.

Published on 29 September 2026 · Reviewed on 6 October 2026

The safe haven shifts to the dollar as Türkiye's gold buffer thins

Open the brief ›
  • Scenario

    Falsified

    The published claim

    H1 · 50% — The rates channel stays dominant: headline PCE on 30 September comes in above the 3.78% forecast and Hormuz does not reopen; the probability of an October Fed hike stays above 70%, the dollar keeps its safe-haven role and gold stays under pressure.

    What happened

    The trigger did not fire. BEA data on 30 September put August headline PCE at 3.4% year on year and core at 3.0%, against expectations of 3.7%. The CME FedWatch probability of a 28 October hike fell from 51% to 35%, then to 20% after the 2 October payrolls report showed just 29,000 jobs. The core of our main scenario, a hike probability above 70%, fell to below a quarter within a week.

    Evidence: 30 September 2026 · Softer US inflation erodes the odds of an October rate rise
  • Scenario

    Partly borne out

    The published claim

    H2 · 30% — Data brings relief, tension persists: PCE comes in below expectations and oil stays near $105; hike odds recede, yields ease somewhat, gold partly recovers, and Türkiye sees limited relief in 2-year yields and CDS.

    What happened

    The rates leg held. PCE came in below expectations and hike odds fell to 35%, then to 20% after the payrolls data; the US 10-year yield eased from 5.29% on 30 September to 5.18% on 2 October. The Türkiye leg failed: 5-year CDS widened from 248.08 on 25 September to 255.69 basis points on 2 October. Nor did oil stay near $105; the December contract closed at $98.03 on 30 September and $100.32 on 5 October.

    Evidence: 2 October 2026 · US economy adds only 29,000 jobs in September
  • Scenario

    Falsified

    The published claim

    H3 · 20% — The supply shock hardens again: a new attack hits Yanbu or the Red Sea, Brent moves sustainably above $108, and currencies and CDS of energy importers widen.

    What happened

    The attack leg partly materialised. The Houthis fired missiles at southern Saudi Arabia on 2–3 October and claimed a hit on an Aramco facility in Riyadh (Saudi Arabia did not confirm it). The price leg did not: front-month December Brent closed at $102.31 on 1 October, $102.25 on 2 October and $100.32 on 5 October, never approaching $108.

    Evidence: Investing.com — Brent futures historical data
  • Falsifier

    Still open

    The published claim

    The converging-signal reading (South Asia's energy importers enter winter with supply and financing gaps) weakens if Qatar lifts force majeure before December or the rupee returns below 95 after the Reserve Bank of India's 5–7 October meeting.

    What happened

    The verdict is open. The Reserve Bank of India's decision is due on 7 October, and as of 6 October we have no published record of Qatar lifting force majeure. New data supports the reading: a deadlock between the IMF and Pakistan over fuel subsidies and gas debt emerged on 1 October, and Pakistan's trade deficit widened by 15% in the first quarter.

  • Watched indicator

    Partly borne out

    The published claim

    Primary item to watch: US PCE data on 30 September. If the headline comes in above 3.78%, the probability of an October hike stays above 70% and dollar yields keep their safe-haven role.

    What happened

    The data was the week's most decisive event, but it worked in the opposite direction to the one we flagged. The headline, at 3.4%, came in well below expectations, and hike odds fell from 51% to 35%. Our warning about the methodology change was written for core; the deviation exceeded 0.3 points on the headline too. Our reading foregrounded the hawkish branch and left the dovish branch in the detail.

    Evidence: 30 September 2026 · Softer US inflation erodes the odds of an October rate rise
  • Watched indicator

    Confirmed

    The published claim

    Secondary item to watch: CBRT reserve data and the fuel tax decision, both due on 1 October. Together they will show Türkiye's external buffer and its domestic inflation path on the same day.

    What happened

    Both arrived on 1 October. Net reserves fell by 2.5 billion dollars to 53.4 billion; net reserves excluding swaps shrank by 3.2 billion dollars in a week to below 40 billion, and gold reserves lost 1.957 billion dollars in value. The same day the sliding-scale mechanism ended and the special consumption tax on petrol was spread over 3 months.

    Evidence: 1 October 2026 · CBRT net reserves excluding swaps fall below 40 billion dollars

What we learned

We got the main scenario (50%) wrong. We assumed the oil shock would push the Fed to hike; instead PCE came in below expectations at 3.4% on 30 September, and the payrolls data cut hike odds to 20%. H2, which proved right, only half held: US rates eased, but Türkiye's CDS widened to 255.69 basis points. In other words, relief originating in the US did not pass through to a Türkiye grappling with a domestic funding crunch and a reserve drain. From now on we will not write scenarios that tie Türkiye's risk premium to the Fed channel alone; domestic fragility will be added as a separate trigger.

Published on 30 September 2026 · Reviewed on 6 October 2026

The tightening wave widens as oil measurement blurs

Open the brief ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Tightening broadens, oil stays high: PCE beats expectations and Hormuz flows stay at 7–10 million barrels a day; the Fed and BoE hike in October–November, the December Brent contract holds in a $95–105 band and the dollar strengthens.

    What happened

    The oil leg held; the rates leg did not. December Brent closed at $98.03 on 30 September, $102.31 on 1 October, $102.25 on 2 October and $100.32 on 5 October, staying inside the band. But headline PCE came in below expectations at 3.4%, and after the 2 October payrolls report showed just 29,000 jobs, the odds of an October Fed hike fell to 20%. The 10-year yield eased from 5.29% to 5.18%.

    Evidence: 2 October 2026 · US economy adds only 29,000 jobs in September
  • Scenario

    Falsified

    The published claim

    H2 · 30% — The Qatar channel yields an interim arrangement: the US responds positively and in writing to Iran's new proposal, a phased reopening of the Strait begins and the oil risk premium recedes.

    What happened

    The opposite happened. On 1 October the US added Iran's rail, automotive and steel sectors to its sanctions list, and on 2 October it announced it was sending a third aircraft carrier to the region. No positive written response was published. Under blockade, Iran could not load a single barrel of crude in September, and no phased reopening began.

    Evidence: 1 October 2026 · US sanctions Iran's railway and car industries
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — Talks collapse and the data darkens: one side shuts the Qatar channel or attacks on ships in the Strait pick up again; front-month Brent climbs back above $105.

    What happened

    The attack leg of the trigger fired. Three tankers were hit on 29 September, UKMTO reported another tanker struck on 1 October, and JMIC confirmed 4 attacks in 96 hours. The price effect did not follow: December Brent closed no higher than $102.31 and fell to $100.32 on 5 October. Attacks are no longer lifting the front month above $105.

    Evidence: 1 October 2026 · Another tanker struck in the Strait of Hormuz catches fire
  • Falsifier

    Still open

    The published claim

    The converging-signal reading (the metrics decision-makers rely on lost credibility in four places in the same week) weakens if IMF PortWatch data for the last week of September narrows the gap between Kpler and the official US figure to below 2 million barrels a day.

    What happened

    The verdict is open. As of 6 October the PortWatch series runs only to 27 September and counts transits rather than barrels (4 transits on 26 September, 1 on 27 September), so the falsifier cannot be tested directly against it. New data supports the reading: on 3 October Iraq's Oil Ministry said the circulating figure of 4 million barrels referred to a single day's loadings.

  • Watched indicator

    Confirmed

    The published claim

    Primary item to watch: US PCE data on 30 September and Brent's roll to the December contract on 1 October. A headline above 3.7% lifts October hike odds above 68%; a December contract opening below $100 should be read as a calendar effect, not supply relief.

    What happened

    Both items shaped the week. Headline PCE came in at 3.4%; with the threshold not crossed, hike odds fell to 35% rather than rising to 68%. The Brent warning proved right: the November contract expired at $103.50 on 30 September while December stood at $98.03. December rose to $102.31 the next day, so the sub-$100 print was not supply relief.

    Evidence: 30 September 2026 · Expiring November Brent settles $5 above the December contract
  • Watched indicator

    Confirmed

    The published claim

    Secondary item to watch: Türkiye's September inflation on 5 October. A monthly figure above the 2.18% expectation would weaken expectations of a rate cut on 22 October.

    What happened

    On 5 October TurkStat put September inflation at 1.84% on the month and 29.73% on the year; the annual rate fell below 30% for the first time since November 2021. With the figure below the threshold, the scenario that would have weakened expectations of a 22 October cut did not materialise. The Bloomberg HT survey had expected 2.20% on the month and 30.30% on the year.

    Evidence: Bloomberg HT — September inflation figures released

What we learned

Our oil reading worked well: we flagged the contract roll in advance, and December Brent stayed within the $95–105 band. Where we went wrong was rates. H1 (50%) assumed PCE would beat expectations and the Fed would hike in October; the data came in below, and hike odds fell to 20%. The 30% we gave H2 was also too high, as within a week the US moved the other way with sanctions and a military build-up. Attacks increased, but the price did not break $105, showing that Hormuz attacks now show up in insurance and transshipment costs rather than in price. We built the falsifier on a unit the PortWatch series we had does not measure; from now on falsifiers will be tied to the unit of the series in hand.

Published on 27 September 2026 · Reviewed on 4 October 2026

The spread draws a border: the Hormuz disruption is now priced outside the US, and Brent-linked importers pay for the $11.91 gap

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Escorted lock, spread above $10: the standoff and blockade continue and the diesel ban debate stays on the table. Brent above $100, WTI in the $90s, the Brent–WTI spread in a $10–14 band, diesel in Türkiye at 90–100 lira.

    What happened

    The oil legs largely held. November Brent expired on 30 September at $103.50 and WTI closed at $90.42, putting the same-month spread at $13.08, inside the band. December Brent, however, stood at $98.03 that day, so the $100 condition briefly broke at the contract roll. WTI dipped just below the $90s to $89.38 on 29 September. We could not verify the diesel leg against October prices; the diesel ban had not become a formal decision as of 28 September.

    Evidence: 30 September 2026 · Expiring November Brent settles $5 above the December contract
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Phased package, spread narrows: a scaled-down offer via Qatar or Oman is accepted, Hormuz transits exceed 20 a day and the ban is shelved; the spread falls below $8 and Brent below $100.

    What happened

    The channel opened but the offer was not accepted. Iran passed its new proposal via Qatar on 29 September; Washington held to its nuclear condition and announced new sanctions on 1 October. Windward counted 24 transits on 27 September, but the number fell back to 16 and 17 on 28 and 29 September. The same-month spread stayed between $12.68 and $13.08 on 28–30 September and never fell below $8.

    Evidence: 29 September 2026 · Iran gives Qatar new Hormuz proposal as Washington insists on nuclear condition
  • Scenario

    Falsified

    The published claim

    H3 · 15% — Diesel wall and a new attack: the US curbs diesel exports, or a new facility or tanker attack occurs around the Red Sea and Yanbu; the spread exceeds $14 and Brent returns above $108.

    What happened

    The outcome leg did not materialise: the same-month spread peaked at $13.08 on 30 September, and Brent's highest close was $105.28 on 28 September. The trigger is disputed. The Houthis hit a power station in Medina on 29 September and fired 5 missiles at southern Saudi Arabia on 1–2 October. Yet by 4 October there had been no confirmed attack on a tanker or oil facility in the Red Sea. Nor had the US issued a formal diesel export curb as of 28 September.

    Evidence: 2 October 2026 · After Medina, the Houthis keep firing missiles at southern Saudi Arabia
  • Causal chain

    Still open

    The published claim

    Step 1: Because of freight and a possible ban, US crude and diesel exports decline; product supply tightens in Europe and the Mediterranean, and the CIF Mediterranean diesel price rises faster than Brent.

    What happened

    September export data were not available as of 4 October. The only data we have point the other way: according to S&P Global, US diesel exports hit a record 1.6 million barrels a day in August. EIA data of 30 September showed distillate stocks down 2.3 million barrels at 105.2 million. We did not measure the CIF Mediterranean diesel price in this 7-day window; the verdict awaits September export data.

    Evidence: Cyprus Shipping News (S&P Global) — Trump to consider diesel export restrictions
  • Causal chain

    Partly borne out

    The published claim

    Step 3: Costly diesel and high Brent pass into the energy bill; energy imports keep rising against 2025 and the current account burden grows; the indicator to watch is TÜİK's September–October energy imports.

    What happened

    The first data point confirmed the direction: August energy imports rose from about $4.83 billion a year earlier to $6.464 billion. Crude import volumes fell 1.1%, so the increase came from price, not volume. But the $6.464 billion August figure does not separate the report's diesel channel from the Brent effect. The September energy bill was not broken out in the flash data of 3 October.

    Evidence: 30 September 2026 · Energy bill widens Türkiye's August trade deficit by 22%
  • Watched indicator

    Still open

    The published claim

    Threshold: dollar/lira above 50 is the zone where, with Brent above $100, the currency channel carries a second wave of increases to the pump price.

    What happened

    The threshold was not crossed: dollar/lira was 48.933 on 25 September and 49.034 on 1 October. The lira weakened about 0.2% in 4 business days; at that pace the 50 mark arrives in months, not weeks. Because the report set no deadline, the verdict on the 50 threshold remains open.

What we learned

We read correctly that the spread would stay wide: the same-month gap was $12.68–13.08 on 28–30 September. But we could not prove why. We attributed about $4 of the spread to the diesel ban; the ban had not become a decision as of 28 September, yet the spread still reached $13. The first link of the chain said exports would decline, but US diesel exports were at a record 1.6 million barrels a day in August. We also made a measurement error: on 1 October Brent rolled to December while WTI stayed in November, and the mismatched comparison showed a spread of $9.44. From now on: (1) the spread is always calculated on the same delivery month; (2) a claim that splits the spread into components is written with a separate series for each; (3) the Türkiye leg does not enter the chain unless the CIF Mediterranean diesel series is tracked.

Published on 28 September 2026 · Reviewed on 4 October 2026

Both exit gates under pressure as markets price a lasting disruption

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Both gates stay narrow: no round of talks is announced, daily Hormuz transits stay below 18 ships and Houthi attacks continue. Saudi exports remain tied to the low-throughput East-West pipeline, Brent trades at $100–108 and the spread above $10.

    What happened

    The price leg held: November Brent closed at $105.28, $102.35 and $103.50 on 28–30 September, and the December contract stood at $102.25 on 2 October. No date for talks was announced by 4 October. The Houthis hit a power station in Medina on 29 September, and 2 missiles and 4 drones were shot down on 1 October. The transit leg was mixed: Windward counted 24 transits on 27 September, 16 on 28 September and 17 on 29 September. The effect leg proved wrong; Saudi exports returned not to the East-West pipeline but to Hormuz, reaching 2.58 million barrels a day in September.

    Evidence: 2 October 2026 · After Medina, the Houthis keep firing missiles at southern Saudi Arabia
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Talks reopen: Washington and Tehran announce a date for a new round this week or in early October, and a written step on the blockade or sanctions follows. Brent falls below $100 and the spread narrows.

    What happened

    The trigger did not come. Iran passed its new proposal via Qatar on 29 September, but no date for a round had been announced by 4 October. The written step came in the opposite direction: on 1 October the US Treasury added Iran's rail and automotive sectors to its sanctions list. Brent's December contract touched $97.09 and $98.03 on 29–30 September, but that dip came from the contract roll, not diplomacy. The December contract was back at $102.25 on 2 October.

    Evidence: 1 October 2026 · US sanctions Iran's railway and car industries
  • Scenario

    Falsified

    The published claim

    H3 · 20% — The Red Sea gate narrows too: damage at Yanbu is confirmed or East-West pipeline throughput falls again, Bab el-Mandeb transits decline, and Brent settles above $108.

    What happened

    The opposite happened: on 28 September Saudi Arabia resumed loadings at Yanbu that had been halted for 17 days. Brent touched $108.83 that day but retreated on the news to close at $105.28. No close above $108 occurred up to 4 October. Bab el-Mandeb flows fell from 5.45 million to 2.56 million barrels a day in September, but that drop came from Saudi cargoes shifting to Hormuz.

    Evidence: 28 September 2026 · Brent retreats from its intraday high as Yanbu news trims the premium
  • Causal chain

    Falsified

    The published claim

    Step 1: While the diplomatic gate stays shut, Saudi exports load onto the low-throughput East-West pipeline. The line opened on 22 September, Aramco is trying to raise throughput to about 4 million barrels a day, and visible Hormuz transits stand at 9 ships.

    What happened

    The diplomatic gate stayed shut until 4 October, but the cargo flowed the other way. According to Kpler data, Saudi shipments through Hormuz rose to 2.58 million barrels a day in September, reaching 43% of oil traffic through the strait. As of 30 September the East-West pipeline was carrying 2.0–2.65 million barrels a day against a capacity of 7 million. We tied the first of the chain's 3 links to the wrong route.

    Evidence: 29 September 2026 · Saudi oil swings back to Hormuz as Bab el-Mandeb flows halve
  • Watched indicator

    Confirmed

    The published claim

    Watch whether the Brent–WTI spread falls from $11.91 on 25 September to below $10; if the spread closes, the seaborne premium is unwinding.

    What happened

    On same-month contracts the spread stayed wide: the gap between November Brent and November WTI was $12.68 on 28 September, $12.97 on 29 September and $13.08 on 30 September. After 1 October Brent rolled to December while WTI stayed in November. On that mismatched comparison the spread was $9.44 on 1 October and $11.14 on 2 October. The one-day dip below $10 coincided with the contract roll and did not last.

    Evidence: Investing.com — WTI crude oil futures historical data
  • Watched indicator

    Falsified

    The published claim

    Threshold: daily Hormuz transits settling above 18 ships, the 10-day average of 24 September, would be the first sign that an opening is reaching the water.

    What happened

    The threshold was crossed without an opening: Windward counted 24 transits on 27 September, 14 of them made with AIS switched off. We had built the threshold on Reuters' series, which showed 9 transits on 24 September, but the breach appeared in a different series. PortWatch recorded 1 transit on 20 September for the same period. There was no date for talks by 4 October; the count measured dark transits, not an opening.

    Evidence: 28 September 2026 · More than half of Hormuz transits now happen with transponders off

What we learned

We were right on price and wrong on mechanism: Brent stayed in the $100–108 band for 6 sessions and the same-month spread widened to $12.68–13.08. But the two-gate picture at the report's core did not hold: Yanbu resumed loadings on 28 September and Saudi cargoes shifted to Hormuz at 2.58 million barrels a day. We built the 18-ship Hormuz threshold on a single AIS series; 14 of the 24 transits on 27 September were dark. From now on 3 rules apply. (1) Hormuz thresholds are written as barrel flows from at least 2 series, not ship counts. (2) Every Brent threshold is written with its contract month, because the 1 October roll moved the headline by about $5. (3) A route claim does not enter the first link of a chain without Kpler-type flow data.

Published on 28 September 2026 · Reviewed on 3 October 2026

The oil shock lands in the week Türkiye's tax shield expires

Open the brief ›
  • Watched indicator

    Confirmed

    The published claim

    Primary item to watch: the fuel tax decision due on 1 October. If the sliding-scale mechanism is lifted entirely, petrol could rise by roughly 12.48 lira a litre; if a phased decision emerges, the burden shifts to the budget.

    What happened

    The decision appeared in the Official Gazette dated 1 October, and the second of the two outcomes materialised. The sliding-scale mechanism ended, but the special consumption tax (ÖTV) on petrol returns over 3 months rather than in one step. It is 7.90 lira in October, 11.36 lira in November and 14.8277 lira in December. October's first step adds roughly 4.15 lira at the pump; a one-off return would have meant 12.47 lira. The ÖTV on LPG rose from 9.22 lira to 11.38 lira per kilogram. The remaining difference stays in the budget as lost tax revenue from October to December.

    Evidence: 1 October 2026 · Türkiye ends its sliding-scale fuel tax and phases in petrol levies
  • Scenario

    Confirmed

    The published claim

    H1 · 50% — The shield is lifted gradually: before 1 October a decision phases in the ÖTV increase or partly extends the sliding scale; no date is announced for US–Iran talks.

    What happened

    Both triggers arrived, but the timing slipped by a day. The phasing decision was published not before 1 October but in the Official Gazette dated 1 October, spreading the petrol ÖTV over 3 months. No official date for US–Iran talks was announced. On the contrary, on 1 October the US Treasury added Iran's automotive, rail and steel sectors to its sanctions list.

    Evidence: 1 October 2026 · Türkiye ends its sliding-scale fuel tax and phases in petrol levies
  • Scenario

    Falsified

    The published claim

    H2 · 30% — The shield is lifted entirely: on 1 October the regulation stays unchanged and the calculated amount is applied in full; Brent stays above 100 dollars.

    What happened

    The regulation was not applied unchanged; the one-off petrol increase of roughly 12.47 lira a litre was split over 3 months. The Brent leg did hold: the November contract closed at 103.50 dollars on 30 September. However, the benchmark December contract stood at 98.03 dollars the same day, so the 100-dollar threshold was missed at the contract roll.

    Evidence: 30 September 2026 · Expiring November Brent settles $5 above the December contract
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — Diplomacy takes the lead: an official date for US–Iran talks is announced and transits through the Strait of Hormuz begin to rise.

    What happened

    The scenario's two legs diverged. No talks date was announced; on 1 October the US stepped up pressure with new sanctions decisions. The transit leg, however, materialised without diplomacy. Windward data showed 17 transits on 29 September, up from 16 the day before. The rise continued even though 3 ships were struck that same day. We had tied transits to diplomacy; the increase came without it.

    Evidence: 29 September 2026 · Tankers are hit in Hormuz, yet transits still rise to 17
  • Falsifier

    Still open

    The published claim

    The converging-signal reading holds that crisis costs are passing into budget and regulatory decisions. It is disproved if three things occur in the same period: a Turkish decision extending or phasing the sliding scale, the Reserve Bank of India holding rates in October, and the US 10-year yield falling below 5%.

    What happened

    The verdict is open; the Reserve Bank of India decides on 7 October. One of the 3 legs has occurred: Türkiye spread the petrol ÖTV over 3 months. The US 10-year yield did not fall below 5%; it stood at 5.29% on 30 September (FRED). We built this falsifier badly. Because the phasing decision writes the lost tax revenue into the budget, it does not disprove the reading; it supports it.

What we learned

The reading on Türkiye proved correct. The 1 October decision was the week's most decisive data point, and the burden shifted to the budget in phases. We were wrong in two places. We tied Hormuz transits to diplomacy, yet transits rose to 17 on 29 September with no talks date. We also tied the falsifier's first leg to the same fact that served as evidence for the thesis; phasing could not count as both falsifier and evidence. From now on, a falsifier will be tied to a fact that shows the opposite of the thesis.

Published on 25 September 2026 · Reviewed on 2 October 2026

Global term premium reaches Türkiye: the 2-year yield tops the 37% policy rate as net reserves shed 6.4 billion dollars in a week

Open the report ›
  • Causal chain

    Falsified

    The published claim

    First link in the chain: high dollar yields would keep the CBRT drawing on reserves, and the weekly data due on 1 October would show net reserves falling by more than 5 billion dollars again.

    What happened

    The direction was right but the size was wrong. Data for the week of 25 September, released on 1 October, showed net reserves falling by 2.5 billion dollars to 53.4 billion dollars. That is half the 5 billion dollar threshold we set. The erosion did not stop, however. Gross reserves fell for a fifth week, by 3.2 billion dollars to 171.2 billion dollars. Net reserves excluding swaps dropped to 39.9 billion dollars, a loss of 16 billion dollars in a month.

    Evidence: 1 October 2026 · CBRT net reserves excluding swaps fall below 40 billion dollars
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — Controlled defence with tight liquidity: US yields stay high, and the CBRT tightens funding without touching the policy rate, slowing its use of reserves.

    What happened

    Two legs of the scenario held, but its core mechanism did not. The US 30-year yield stayed above 5.5% on 29 September, the policy rate was unchanged at 37.00%, and the weekly reserve loss slowed from 6.4 billion dollars to 2.5 billion dollars. On 1 October, however, the CBRT did not tighten but eased. It raised the SME loan growth cap from 4.5% to 5% and cut the reserve requirement blocking ratio for large banks from 40% to 35%.

    Evidence: 1 October 2026 · CBRT raises the growth cap on SME lending to 5%
  • Scenario

    Still open

    The published claim

    H3 · 20% — Forced into a rate move: the US 30-year yield tops 5.5%, net reserves fall below 50 billion dollars and the dollar/lira passes 50, so the CBRT has to raise the policy rate.

    What happened

    Only one of the three triggers arrived. The US 30-year yield rose above 5.5% and closed at 5.61% on 1 October. Net reserves, at 53.4 billion dollars, remained above 50 billion dollars, and the dollar/lira stayed below 50 at 49.12 on 1 October. The rate decision will be taken at the MPC meeting on 22 October, so the verdict remains open.

  • Scenario

    Still open

    The published claim

    H2 · 30% — Global relief: demand at US auctions recovers, the 10-year yield returns below 5%, Brent falls below 100 dollars, and reserve losses stop.

    What happened

    There was no sign of relief within the week. The US 10-year yield closed at 5.24% on 1 October. Brent's front-month contract dropped to 98.03 dollars on 30 September with the contract rollover, but returned to 102.31 dollars on 1 October. Reserve losses slowed but did not stop. Because the scenario did not specify a time frame, the verdict remains open.

What we learned

Where we went wrong was the CBRT's reaction function. We assumed the bank would tighten funding as reserves eroded. Instead, once the fund crisis reached the banking sector through 3 investment banks on 30 September, the priority became keeping the credit and liquidity channel open. The lesson: in Türkiye scenarios we will write a financial stability shock as a separate trigger. Tying the reserve threshold to the level of net reserves excluding swaps, rather than to the weekly change, is also a sturdier measure.

Published on 25 September 2026 · Reviewed on 1 October 2026

Long end at 1993 high, Türkiye's short-term yields overtake the policy rate, deals deferred to January

Open the brief ›
  • Watched indicator

    Partly borne out

    The published claim

    Secondary watch: Iran's deadline expiring on 28–29 September. The US response to its 60-day ceasefire roadmap would decide whether transits in the Strait of Hormuz rose from 10 ships a day and which way Brent moved from $106.60.

    What happened

    The response came before the deadline and it was negative: on 26 September Trump rejected Iran's 7-day Hormuz plan. Tehran passed on a new proposal through Qatar on 29 September, and Washington held firm on its nuclear condition. Brent, however, did not react as we expected. It rose above $106 in Asian trading on 28 September on news of the rejection, but closed at $102.35 on 29 September, $4.25 below the $106.60 of 24 September. The indicator's question was answered, but it was not what set the direction of prices.

    Evidence: 29 September 2026 · Iran gives Qatar new Hormuz proposal as Washington insists on nuclear condition
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Hormuz opens and rate pressure eases. The US accepts Iran's 60-day ceasefire roadmap and Hormuz reopens in stages. Brent falls below $100, US yields return below 5% and Türkiye's CDS eases.

    What happened

    The trigger did not come: Trump rejected Iran's plan on 26 September. Brent did not fall below $100; its lowest close between 25 and 29 September was $102.35 on 29 September. The US 10-year yield hit a 19-year high on 28 September, and the 30-year stayed above 5.5% on 29 September. Türkiye's CDS did not ease, closing at 248.08 on 25 September and 249.98 on 28 September.

    Evidence: 26 September 2026 · Trump rejects Iran's plan to reopen Hormuz in 7 days; Tehran had set conditions on the blockade, oil sanctions and around 12 billion dollars in frozen assets
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — Double rupture. Iran's deadline goes unanswered, a second wave of attacks hits the Red Sea and the US 30-year yield tops 5.5%. Brent rises above $110 and Türkiye's CDS exceeds 250 basis points.

    What happened

    Two of the 3 triggers arrived, but the oil leg worked in the opposite direction. The US 30-year yield rose to 5.50% on 25 September and stayed above 5.5% on 29 September. On 26 September the Houthis launched 2 drones at Riyadh and a ballistic missile at Khamis Mushait. Iran's deadline did not go unanswered; the US rejected it on 26 September. Brent came nowhere near $110, falling from $106.60 on 24 September to $102.35 on 29 September. Türkiye's CDS crossed 250 intraday on 25 September, but closes stayed at 248.08 and 249.98.

    Evidence: 26 September 2026 · Houthis fire drones at Riyadh and a missile at Khamis Mushait after Türkiye–Pakistan–Saudi staff chiefs meet; coalition says 2 missiles, 2 drones downed
  • Scenario

    Still open

    The published claim

    H1 · 50% — The term premium stays high and deferrals continue. The Fed hikes on 28 October, indirect demand at US auctions stays weak, and time-limited extensions continue on the Iran and China files. Oil stays in a $100–110 band.

    What happened

    The verdict is open because the main trigger is the 28 October FOMC decision. As of 30 September, interim findings are consistent with the scenario. Brent closed between $102.35 and $105.28 from 25 to 29 September, inside the band. The US 30-year yield is above 5.5%. On 29 September the market priced a 68% probability of an October hike, against 77.5% on 24 September.

    Evidence: 29 September 2026 · US long-dated yield holds above 5.5% despite Treasury buybacks
  • Watched indicator

    Still open

    The published claim

    Primary watch: the CBRT weekly reserve data at 14:30 on 1 October. If the loss again exceeds $5 billion and the 2-year yield stays above 37%, the market is pricing that non-rate tools will not be enough.

    What happened

    The data will be released at 14:30 on 1 October; this review was written on the morning of the same day. Market calculations put gross reserves down a further $3.4 billion to $171 billion in the week of 25 September. There is no official figure yet for the change in net reserves, so no verdict has been given.

  • Falsifier

    Still open

    The published claim

    The reading of time-limited deferrals fails if, before 10 January, the US and China sign a lasting tariff and rare earths package, or Iran and the US reach a Hormuz agreement with no time limit.

    What happened

    The deadline is 10 January 2027, so the verdict is open. As of 30 September neither condition had been met. The US rejected Iran's 7-day plan on 26 September, and Iran passed on a new proposal on 29 September. That proposal also rests on a time-limited reopening schedule.

What we learned

Where we went wrong was the oil channel. We assumed bad news on the Iran file would push Brent higher. The US rejected the plan on 26 September and the Houthis struck Saudi territory, yet Brent fell $4.25 between 24 and 29 September. In the same week Yanbu resumed loading on 28 September after a 17-day halt, and this supply-side news outweighed the rejection. The scenario should have listed alternative supply routes as triggers, not just the demand side. The second lesson is timing: we tied the primary indicator to the afternoon of the review day. From now on, the scorecard review will be left until the night after the watched data are released.

Published on 24 September 2026 · Reviewed on 30 September 2026

Yields at a 2007 peak, oil back at 103 dollars, the strait tied to 7 conditions

Open the brief ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — High rates, partial supply: the 7-year auction comes in weak, the October hike probability stays around 70%, Hormuz transits remain in single digits and the East-West pipeline runs at low flow. Brent fluctuates above 100 dollars, and in Türkiye the currency and reserve channel works rather than the CDS.

    What happened

    The main line held, but 2 legs did not. The 7-year auction closed weak on 24 September, at 5.085% with 57.2% indirect bidding. The October hike probability was 68% on 29 September. Brent fell from 106.60 dollars on 24 September to 102.35 dollars on 29 September but stayed above 100. Yanbu resumed loading on 28 September; full capacity needs 6–8 weeks. First miss: Windward counted 24 transits on 27 September, 14 with transponders off, so transits left single digits. Second miss: Türkiye's CDS rose from 245.17 on 24 September to 249.98 on 28 September, while USD/TRY rose only 0.15% on 25 September. The reserve channel did work: gross reserves fell 3.4 billion dollars in the week of 25 September.

    Evidence: 29 September 2026 · Markets price 68% odds of October Fed hike before PCE data
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Breathing space: the 7-year auction closes well, a timetable for extending the ceasefire is announced at the summit and Hormuz transits exceed 10 for several days running. Part of the jump in the 5-year yield is reversed, Brent approaches levels below 100 dollars and the October hike probability recedes.

    What happened

    The scenario did not materialise, and 2 of its 3 effect legs went the opposite way. The 7-year auction closed on 24 September at 5.085%, the highest rate since April 1993. The bid-to-cover ratio fell from 2.50 to 2.42. Yields were not given back: the 10-year rose from 5.18% on 24 September to 5.244% on 28 September, its highest since July 2007. Brent did not approach levels below 100 dollars; the lowest close in the review window was 102.35 dollars on 29 September. The only partial match is in the probability: the October hike fell from 77.5% on 24 September to 68% on 29 September. Windward's 24 transits on 27 September are a single day. Our records show no run of consecutive days above 10.

    Evidence: 28 September 2026 · US ten-year Treasury yield climbs to a 19-year high
  • Scenario

    Falsified

    The published claim

    H3 · 20% — Double shock: new attacks on ships continue at Hormuz, Libya declares force majeure and the US 30-year yield passes 5.5%. The Fed's hike timetable is brought forward, oil, diesel and freight rise, and Türkiye's CDS tests the 250 basis point threshold.

    What happened

    Only 1 of the 3 triggers arrived, so the scenario was falsified as written. The US 30-year yield rose to 5.50% on 25 September, 5.55% on 28 September and 5.57% on 29 September. Libya did not declare force majeure; on the contrary, the NOC reopened the Sharara valve on 26 September after 6 days. Our records for 25–29 September show no new attack on a ship at Hormuz. Oil did not rise: Brent fell from 106.60 dollars to 102.35 dollars. Even so, the market signal we had tied to the double shock arrived without the triggers. According to Bloomberg HT, Türkiye's CDS passed 250 on 25 September for the first time since 27 May. The closes were 248.08 on 25 September and 249.98 on 28 September.

    Evidence: 25 September 2026 · Türkiye's CDS tops 250 for the first time since 27 May: as the $18 billion liquidation of 131 funds proceeds, Erdoğan says no systemic risk; 45 arrested
  • Watched indicator

    Partly borne out

    The published claim

    Primary watch: the 7-year auction on 24 September and the 30-year yield. A tail like the 5-year's and indirect bidding below 60% would mean the demand problem is not a one-day event. A 30-year yield passing 5.39% and nearing 5.5% while the 2-year holds steady would show tightening moving into the term premium.

    What happened

    The indicator half worked; 2 of 4 conditions held. Indirect bidding, at 57.2%, was below the 60% threshold, against an average of 64.6%. The tail, however, did not resemble the 3.1 basis points of the 5-year and stayed at 0.7 basis points. The 30-year yield passed 5.39% on 24 September at 5.47% and reached 5.57% on 29 September. The other half of the condition failed: the 2-year yield did not hold steady, rising from 4.85% on 23 September to 4.93% on 29 September. The 30-year rose 17 basis points and the 2-year 8 basis points. The term premium and policy expectations worked together.

    Evidence: 24 September 2026 · US Treasury sells 44 billion dollars of 7-year notes at 5.085%: highest yield since April 1993 as indirect demand falls to 57.2%
  • Watched indicator

    Confirmed

    The published claim

    Secondary watch: the CBRT's official reserve data at 14:30 on 24 September. Confirmation of the 174.5 billion dollars in the market calculation and of the 13.9 billion dollar loss over 4 weeks would show the cost of defending the currency with USD/TRY at 48.84.

    What happened

    The official data confirmed the calculation on both figures. On 24 September the CBRT published total reserves of 174.405 billion dollars for the week of 18 September, a weekly fall of 4.319 billion dollars. Bloomberg HT put the total 4-week loss at about 14 billion dollars. Net reserves fell by 6.41 billion dollars to 55.84 billion dollars, and net reserves excluding swaps fell to 43.1 billion dollars. The erosion continued: according to the market calculation, gross reserves fell a further 3.4 billion dollars to 171 billion dollars in the week of 25 September. Official confirmation of this last figure is due on 1 October.

    Evidence: 24 September 2026 · CBRT official data: in the week of 18 September gross reserves fell by 4.3 billion dollars to 174.4 billion, net reserves by 6.4 billion to 55.8 billion
  • Falsifier

    Still open

    The published claim

    The spare capacity reading is refuted if 3 conditions are met together by mid-October. Libya's NOC announces that the Sharara valve and the Zawiya refinery are operating, the East-West pipeline approaches 4 million barrels a day, and at least 2 of Russia's 3 halted major diesel refineries return to production.

    What happened

    The verdict is open because the deadline is mid-October. As of 30 September none of the 3 conditions has been fully met. On 26 September the NOC announced that valve number 7 on the Sharara line had been opened. The return of the unit shut at Zawiya that same day could not be verified. Loading at Yanbu began on 28 September after a 17-day pause, and a return to 4 million barrels a day could take 6–8 weeks. The Russian side is moving the other way: Perm and Novoshakhtinsk halted on 25 September and Ilsky on 26 September. Putin classified refinery throughput data on 29 September, which makes confirming 2 plants under the third condition harder.

    Evidence: 26 September 2026 · Libya's Sharara–Zawiya pipeline reopens after six days

What we learned

We were wrong on the Türkiye channel. On 24 September we wrote that the currency and reserves, not the risk premium, would be decisive; the CDS rose from 239.61 to 249.98 in 4 trading days. We had tied the 250 threshold to the double-shock branch, rated at 20%, yet it was tested before 2 of that branch's 3 triggers arrived. The cause was a domestic shock we had not counted: the liquidation of 131 funds worth 18.3 billion dollars. Our second error was relying on one Hormuz series; Windward counted 24 transits on 27 September, 14 with transponders off. Third, we assumed a steady 2-year yield, and it rose 8 basis points. The method changes in 3 ways. Market thresholds will carry 2 trigger paths each, not one scenario. Domestic financial stability becomes 1 separate line in the Türkiye reading. Strait traffic will use at least 2 series, AIS and satellite counts.

Published on 23 September 2026 · Reviewed on 29 September 2026

Risk premium moves to the core: CDS widened in all 14 of the 14 countries reporting on 21 September, and France overtook Italy

Open the report ›
  • Scenario

    Confirmed

    The published claim

    H1 · 50% — Synchronised tightening, premiums in a high band: energy stays high, developed and emerging-market premiums settle at 21 September levels, and France stays above Italy.

    What happened

    The direction held, but the level proved harsher than expected. Premiums did not stabilise on 21 September; they widened further. France's 5-year CDS rose from 47.19 on 21 September to 57.42 basis points on 24 September. Italy stayed at 45.12 the same day, so France remained above Italy. Germany rose from 8.49 to 9.50, and Türkiye from 242.43 to 245.17.

    Evidence: Investing.com — France CDS 5 Years USD
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Energy retreats, premiums narrow: Brent settles well below 100 dollars, part of the 21 September widening is reversed, and Türkiye's CDS could return to around 230.

    What happened

    The trigger did not materialise. Instead of falling below 100 dollars, Brent rose to 106.60 dollars on 24 September and again topped 106 dollars at the Asian open on 28 September. Türkiye's CDS did not return to 230; on 25 September it crossed 250 for the first time since 27 May.

    Evidence: 28 September 2026 · Brent tops 106 dollars in Asia as Iran sees no new round
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — A break at the core: the France–Germany spread exceeds 110 basis points and France's CDS passes 50; the ECB is caught between fragmentation risk and tightening.

    What happened

    Half of the trigger materialised. France's 5-year CDS rose above 50, to 55.98 basis points on 23 September and 57.42 on 24 September. Sources diverge on the France–Germany 10-year spread. Calculated from 28 September closes (France 4.78%, Germany 3.63%), it is about 115 basis points, while another provider gives 105.4 for 25 September. We could not verify with a single consistent series that the threshold was crossed. Nor is there any record of the ECB changing policy in response to fragmentation risk.

    Evidence: Investing.com — France CDS 5 Years USD
  • Watched indicator

    Confirmed

    The published claim

    Threshold: Türkiye's CDS exceeding 250 basis points would show the Turkish premium moving beyond the bloc-wide move, with country-specific pressure added.

    What happened

    The threshold was crossed on 25 September, when Türkiye's CDS rose above 250 for the first time since 27 May. The source of country-specific pressure was also clearly visible: the liquidation of 131 funds worth about 18 billion dollars and a widening fund investigation.

    Evidence: 25 September 2026 · Türkiye's CDS tops 250 for the first time since 27 May: as the $18 billion liquidation of 131 funds proceeds, Erdoğan says no systemic risk; 45 arrested
  • Watched indicator

    Confirmed

    The published claim

    Threshold: the US 10-year yield exceeding 5.05% would show the dollar discount rate moving into a new, higher band.

    What happened

    The threshold was crossed on 23 September. After hot PMI data, the 10-year yield rose to 5.10%, its highest level since 2007. The same day the US Treasury sold 5-year notes at 5.033%.

    Evidence: 23 September 2026 · US Treasury sells $70 billion of 5-year notes at 5.033%; after hot PMIs the 10-year yield hits 5.10%, its highest since 2007
  • Causal chain

    Partly borne out

    The published claim

    First link in the chain: tightening is written at the front end; the problem lies in price, not demand. The reading that 'yields are rising because demand is weakening' was flagged as noise.

    What happened

    The price leg held, but we were wrong on demand. The 5-year auction cleared at 5.033% and the 7-year auction on 24 September at 5.085%, the highest rate since April 1993. Yet indirect demand at the 7-year auction fell to 57.2%. In Germany, a 5 billion euro Bobl auction on 23 September drew bids of only 4.53 billion. Weak demand was not noise but a second signal.

    Evidence: 24 September 2026 · US Treasury sells 44 billion dollars of 7-year notes at 5.085%: highest yield since April 1993 as indirect demand falls to 57.2%
  • Watched indicator

    Still open

    The published claim

    Thresholds: Japan's 30-year yield exceeding 4.21% and Germany's 30-year yield exceeding 4.00% would show tightening moving out to the long end.

    What happened

    Neither threshold has yet been crossed. Japan's 30-year yield closed at 4.16% on 25 September and Germany's 30-year yield at 3.91% the same day. The check has not yet matured; the verdict is deferred to the next review.

What we learned

The report read the direction correctly but was wrong in two places. First, we expected premiums to stabilise at 21 September levels; instead the widening continued and Türkiye added a country-specific channel. Second, we treated weak auction demand as noise. The US 7-year and German 5-year auctions showed that demand was also eroding. From now on, when reading auctions we will record price and demand as separate signals.

Published on 23 September 2026 · Reviewed on 28 September 2026

The strait went to the bargaining table, the premium moved to the core

Open the brief ›
  • Scenario

    Falsified

    The published claim

    H1 · 50% — Conditional opening, slow normalisation: Iran confirms the offer at the UN, the US gradually eases the blockade; the 20% penalty regime takes effect but is applied selectively, and transits rise over several weeks.

    What happened

    The central trigger of the scenario we rated most likely did not materialise. The offer was confirmed: Araghchi told the UN General Assembly on 25 September that the Strait of Hormuz would open at the end of day 7, on condition that the blockade and oil sanctions were lifted. But the US did not ease the blockade. Trump rejected the 7-day roadmap on 26 September, and the blockade has turned back 122 ships in total. Transits did not rise: 9 on 24 September and single digits on 25 September. Brent did not fall; it rose to 106.60 dollars on 24 September. Because the negotiating channel has not closed, this branch could still open later; within the review window, however, it was proven wrong.

    Evidence: 26 September 2026 · Trump rejects Iran's plan to reopen Hormuz in 7 days; Tehran had set conditions on the blockade, oil sanctions and around 12 billion dollars in frozen assets
  • Scenario

    Partly borne out

    The published claim

    H2 · 35% — Offer collapses, status quo holds: the offer is not officially confirmed, the blockade continues and the penalty articles become law; transits stay in single digits per day, the Yanbu outlet is loaded up and Brent could rise above 100 dollars again.

    What happened

    The outcome held, but the mechanism did not. Transits stayed in single digits (9 on 24 September), and Brent closed above 100 dollars at 106.60 dollars on 24 September and 104.32 dollars on 25 September. The Yanbu outlet became both a load and a target: the Houthis fired 6 ballistic missiles at Yanbu and Taif on 24 September, and France sent air defences to Yanbu. The first leg of the trigger was wrong, however: the offer was confirmed, with Araghchi presenting it openly at the UN on 25 September. The status quo came not from an Iranian retreat but from the US rejection on 26 September. There is no recorded evidence that the penalty articles passed the plenary.

    Evidence: 24 September 2026 · Houthis fire 6 ballistic missiles at Yanbu and Taif; with the Red Sea end of the East-West pipeline targeted, Brent closes up 3.41% at $106.60
  • Scenario

    Falsified

    The published claim

    H3 · 15% — Tension escalates: a ship is seized under the 20% penalty regime or a new tanker attack occurs, and the US responds; risk also rises on the Red Sea side.

    What happened

    The trigger did not materialise: our records for 24–27 September show no seizure under the penalty regime and no new tanker attack in the Strait of Hormuz. The scenario's impact leg nonetheless arrived without the trigger. Risk rose on the Red Sea side: the Houthis fired missiles and drones at Yanbu on 24 September and at Riyadh and Khamis Mushait on 25–26 September. The escalation thus came from the Yemen front rather than the legal regime in Hormuz, and the scenario, as we wrote it, was proven wrong.

    Evidence: 26 September 2026 · Houthis fire drones at Riyadh and a missile at Khamis Mushait after Türkiye–Pakistan–Saudi staff chiefs meet; coalition says 2 missiles, 2 drones downed
  • Watched indicator

    Confirmed

    The published claim

    Primary watch item: official confirmation of Iran's 7-day offer and the daily count of Hormuz transits. If confirmation arrives and transits exceed 20 for several days in a row, diplomacy is turning into physical flows.

    What happened

    The indicator drew the right distinction: confirmation came, but flows did not. The offer was presented at the UN on 25 September by Foreign Minister Araghchi rather than Pezeshkian. Transits did not reach 20 on any day: 10 on 23 September (later revised to 14), 9 on 24 September and single digits on 25 September, with a 10-day average of 18. Diplomacy stayed on the table without turning into physical flows; Trump's claim of 29 ships could not be verified with independent data.

    Evidence: 25 September 2026 · Iran offers to open Hormuz by the end of day 7 if the US blockade and oil sanctions are lifted; Rubio says no breakthrough, Brent closes 2.1% lower at $104.32
  • Watched indicator

    Confirmed

    The published claim

    Secondary watch item: the length of the tariff truce extension from the 24 September Xi–Trump summit. Under 6 months would mean a brief, conditional timetable; close to 2 years would mean the magnet leverage is working.

    What happened

    The extension proved brief. Bessent announced on 23 September that the truce due to expire on 10 November had been extended to 10 January, only 2 months. The 24 September summit ended without a joint statement and produced no tariff cuts, chip licences or rare earth decisions. China's planned restrictions on 7 additional rare earth metals were also postponed to 10 January 2027. The brief, conditional timetable branch in our reading materialised.

    Evidence: 24 September 2026 · No joint statement at Xi–Trump summit: tariff truce extended by just 2 months from 10 November to 10 January, no new decisions on chips or rare earths
  • Falsifier

    Still open

    The published claim

    The confluence reading (the disruption is being booked as a cost item lasting quarters) is falsified if all three occur in the same period: core PCE falling below 3% for two months running, the 2-year auction yield returning to its August level, and France's premium dropping back below Italy's.

    What happened

    The falsifier has not come due, as it depends on two months of PCE data; the first test is on 30 September. The interim picture supports the reading. The 2-year yield stood at 4.87% on 24 September, well above 4.204% at the August auction, and the 7-year auction closed at 5.085%, the highest rate since 1993. France's 5-year CDS stood at 57.42 on 24 September, still above Italy's 45.12.

    Evidence: 24 September 2026 · US Treasury sells 44 billion dollars of 7-year notes at 5.085%: highest yield since April 1993 as indirect demand falls to 57.2%

What we learned

Where we went wrong was in tying the trigger to the wrong party. We built the split in the Hormuz scenarios on whether Iran would confirm its offer. Yet the offer was confirmed, and the decisive variable was Washington's response. That is why the conditional opening branch, rated at 50%, was proven wrong, while the 35% status quo branch reached the right outcome through the wrong mechanism. The second lesson: we tied the escalation scenario solely to the penalty regime in Hormuz, while prices were carried by the Yanbu attack from the Yemen front. From now on, in negotiation scenarios each side's acceptance step will be written as a separate trigger, and a second front (the Red Sea) will be treated as a separate branch of chokepoint risk.

Published on 23 September 2026 · Reviewed on 28 September 2026

Two channels on Hormuz: a 7-day reopening offer at the table, a fine of 20% of cargo in parliament

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Talks continue, transits stay low: the mediator channel remains open but the US publishes no written plan; visible transits stay in single digits and Brent fluctuates in a 95–105 dollar band.

    What happened

    The political and physical legs held. Araghchi met Witkoff and Kushner on 25 September under Qatari mediation, the US published no written plan, and Trump rejected the offer on 26 September. Transits were 9 on 24 September and in single digits on 25 September. The price band, however, was breached for one day: Brent closed at 106.60 dollars on 24 September and returned to the band at 104.32 dollars on 25 September. The breach came not from a collapse of talks but from the Houthi attack on Yanbu.

    Evidence: 26 September 2026 · Trump rejects Iran's plan to reopen Hormuz in 7 days; Tehran had set conditions on the blockade, oil sanctions and around 12 billion dollars in frozen assets
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Phased opening timetable: the White House announces a plan to ease the blockade or halt operations, and Iran runs its 7-day clock; transits rise above 10 ships and the front end of the crude curve falls markedly.

    What happened

    The trigger did not materialise. The White House announced no plan, Trump rejected Iran's 7-day roadmap on 26 September, and the blockade has turned back 122 ships in total. Transits did not rise durably above 10 (9 on 24 September). Rather than falling, Brent rose from 98.50 dollars on 22 September to 104.32 dollars on 25 September.

    Evidence: 26 September 2026 · Trump rejects Iran's plan to reopen Hormuz in 7 days; Tehran had set conditions on the blockade, oil sanctions and around 12 billion dollars in frozen assets
  • Scenario

    Falsified

    The published claim

    H3 · 15% — Penalty regime kicks in: the bill passes the plenary and Iran enforces its first detention or 20% penalty case; Brent returns above 105 dollars.

    What happened

    The trigger did not materialise: our records for 24–27 September contain no evidence that the bill passed the plenary, nor of a first detention or penalty case. The scenario's price outcome did arrive, but through another channel. Brent rose above 105 to 106.60 dollars on 24 September, driven by the 6 ballistic missiles the Houthis fired at Yanbu and Taif.

    Evidence: 24 September 2026 · Houthis fire 6 ballistic missiles at Yanbu and Taif; with the Red Sea end of the East-West pipeline targeted, Brent closes up 3.41% at $106.60
  • Watched indicator

    Falsified

    The published claim

    Threshold: Brent rising above 105 dollars was to be read as the zone where a collapse of talks, or the start of penalty enforcement, had passed into prices.

    What happened

    The threshold was crossed, but our interpretation proved wrong. Brent closed at 106.60 dollars on 24 September. Talks had not collapsed that day; the next day Araghchi presented the 7-day plan at the UN, and the penalty regime was not being enforced. What pushed the price above the threshold was the Houthi attack on Yanbu, the outlet of the East–West pipeline that bypasses Hormuz. Tying the threshold to a single channel was a mistake.

    Evidence: 24 September 2026 · Houthis fire 6 ballistic missiles at Yanbu and Taif; with the Red Sea end of the East-West pipeline targeted, Brent closes up 3.41% at $106.60
  • Causal chain

    Confirmed

    The published claim

    Chain link 1: unless the White House publishes a written plan on the blockade or operations, visible transits stay in single digits; the offer's condition is tested on the ground.

    What happened

    The link held: no written plan came and transits stayed in single digits. According to Reuters-sourced data, 9 commodity ships passed on 24 September, 8 of them outbound; the 10-day average was 18. The figure for 23 September was revised from 10 to 14, and it should be noted that preliminary data were later revised upwards. The pre-war baseline was about 125 ships a day.

    Evidence: 24 September 2026 · Tracked Hormuz transits fell to 9 on 24 September, 8 of them outbound; 23 September revised from 10 to 14, while The National gave 13 for the same day
  • Causal chain

    Still open

    The published claim

    Chain link 2: if the penalty articles go to the plenary while the opening is delayed, the strait's legal regime changes and the floor of the war-risk premium stays elevated.

    What happened

    The link's horizon was set in weeks and has not yet expired. As of 28 September our records contain no evidence that the bill has been voted on in the plenary; articles 13–15, approved by the committee on 22 September, are pending. We have not collected a new dated value for the war-risk premium.

What we learned

Where we went wrong was in interpreting the price threshold. We framed Brent crossing 105 dollars as a gauge of the Hormuz negotiations or the penalty regime. The threshold was crossed within two days, but the cause was the attack on Yanbu on the Red Sea. The outlet of the pipeline that bypasses the strait is a separate risk channel, and it contaminates the price thresholds of a Hormuz report. The ranking of scenarios, by contrast, proved largely correct: the main scenario held, while the opening and penalty branches did not materialise. From now on, price thresholds will not be tied to a single cause; for each threshold, which channel carried the price will be verified separately against the event record.

Published on 22 September 2026 · Reviewed on 27 September 2026

The benchmark fell, the bill did not

Open the brief ›
  • Scenario

    Confirmed

    The published claim

    H3 · 20% — The demand diagnosis wins: US data confirm that inflation is demand-driven, the Fed moves to a harsher, front-loaded path, the dollar strengthens and emerging-market premia widen indiscriminately.

    What happened

    The scenario to which we gave the lowest probability was the one most fully realised within five days. On 23 September Barr and Collins signalled an additional hike; the probability of an October hike rose from 55% to 70% within the day, and the dollar index moved above 101. On 24 September Paulson said moderate additional tightening might be needed, and jobless claims fell to 197,000. On 25 September the US 30-year yield rose to 5.50%. Premia widened as a bloc: Argentina's country risk reached a six-month high of 609 basis points on 25 September, and Türkiye's CDS rose above 250 for the first time since 27 May. The decisive test of the data leg of the demand diagnosis is the PCE release on 30 September; however, the Fed path and the market leg have already materialised.

    Evidence: 23 September 2026 · Fed's Barr and Collins signal further hikes: October hike odds jump intraday from 55% to 70%, and the dollar index rises above 101
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — The spread stays wide: Strait of Hormuz traffic stays low, Russian refining losses accumulate and the diagnostic split within the Fed persists; while crude eases, refined products and the risk premium stay expensive.

    What happened

    Two of the three triggers held and one did not. Hormuz stayed low: 7 transits on 22 September, between 10 and 14 on 23 September, and 9 on 24 September. Russian refining losses accumulated: Perm and Novoshakhtinsk halted on 25 September. But the split within the Fed did not persist; it converged on the hawkish side. The scenario's price premise also proved wrong: crude did not ease, Brent rose to $106.60 on 24 September and closed at $104.32 on 25 September, roughly 4.3% above the $100.05 of 22 September. The leg about the risk premium staying expensive was confirmed.

    Evidence: 24 September 2026 · Houthis fire 6 ballistic missiles at Yanbu and Taif; with the Red Sea end of the East-West pipeline targeted, Brent closes up 3.41% at $106.60
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Diplomacy turns into traffic: contacts on the margins of the UN General Assembly bear fruit, the weekly transit count in the Strait of Hormuz approaches its pre-war band, and Brent and the product premium retreat together.

    What happened

    The trigger did not materialise within the review window, and the price moved in the opposite direction. Diplomacy stayed on the table but did not turn into traffic: on 25 September Iran offered to open the strait at the end of the seventh day if the blockade and sanctions were lifted, and Rubio said this was not a breakthrough. Transits stayed between 7 and 14 a day; the pre-war baseline, according to sources, is 85 to 125 ships a day. Brent did not retreat; it rose. Because the negotiating track has not closed, this branch may reopen later; however, in the form we wrote it, as an outcome of UN week, it did not materialise.

    Evidence: 25 September 2026 · Iran offers to open Hormuz by the end of day 7 if the US blockade and oil sanctions are lifted; Rubio says no breakthrough, Brent closes 2.1% lower at $104.32
  • Falsifier

    Confirmed

    The published claim

    The sanctions regime is shifting from prohibition to a bargaining chip, and the compliance burden is moving to third-country intermediaries. This reading is disproved if Usmanov or Fridman is put back on the EU list, an equivalent independent monitoring mechanism replaces the UN panel, and no new designations target third-country intermediaries.

    What happened

    None of the three legs of the falsifier materialised; on the contrary, the burden kept shifting to third countries. On 23 September the Central Bank of the UAE barred Bank Melli Iran's 8 branches in the country from trade finance and money transfers involving Iran: the decision was taken not by a sanctioning state but by the regulator of an intermediary country exposed to secondary sanctions risk. Usmanov and Fridman were not put back on the list, and no mechanism replaced the UN panel. The reading stands.

    Evidence: 23 September 2026 · UAE Central Bank bars Bank Melli Iran's 8 branches in the country from trade finance and fund transfers with Iran
  • Watched indicator

    Confirmed

    The published claim

    Secondary watch item: the first common observation day after 21 September on the sovereign risk premium panel; the series for eight countries, Türkiye included, stopped on 18 September, and it was unknown whether they had joined the bloc move.

    What happened

    The question has been answered: Türkiye joined the bloc. In the Investing series the 24 September close was 245.17; according to BloombergHT, on 25 September the CDS exceeded 250 for the first time since 27 May. However, part of Türkiye's widening came through a domestic channel: in the same days the liquidation of 131 funds worth $18 billion and the MASAK investigation were under way. Participation in the bloc move is confirmed; attributing the entire move to the global channel would not be accurate.

    Evidence: 25 September 2026 · Türkiye's CDS tops 250 for the first time since 27 May: as the $18 billion liquidation of 131 funds proceeds, Erdoğan says no systemic risk; 45 arrested

What we learned

Where we went wrong was the ranking. On 22 September we gave diplomacy 30% and the demand diagnosis 20%; five days later the lowest-probability scenario was the most fully realised and the second-ranked one was disproved. The error lies in a single variable: we took the fall in price from $104.82 to $100.05 as evidence of diplomatic expectations and treated the Fed's reaction function as a secondary variable. Yet in the same week Fed speakers converged in one direction, and risk premia were carried by the discount rate, not by oil. H1's price premise stemmed from the same error: we assumed crude would ease, and Brent rose again. From now on, in energy-driven scenarios the direction of central bank speakers and futures-implied rate probabilities will be written as the primary trigger; a single week's price move will not be counted as evidence of diplomacy.

Published on 22 September 2026 · Reviewed on 27 September 2026

Latin America was repriced before it was refinanced: risk premiums widened together on 21 September

Open the report ›
  • Causal chain

    Partly borne out

    The published claim

    Ripple chain: uncertainty over the dollar discount rate widens premia at bloc level, country calendars run into the premium, and if the calendar does not pass cleanly Türkiye's external financing cost rises without any country-specific news. Watch: whether Türkiye's 5-year risk premium rises above 232.80 basis points of 18 September.

    What happened

    The chain's watch indicator crossed the threshold: Türkiye's 5-year CDS rose to 245.17 on 24 September and, according to BloombergHT, above 250 on 25 September for the first time since 27 May; the report's threshold of tr-cds > 250 was also breached. The US 30-year yield reached 5.50% on 25 September. However, the chain's premise of 'without country-specific news' did not hold: in the same days the liquidation of 131 funds worth $18 billion and the MASAK investigation opened a Türkiye-specific channel. The outcome was confirmed, but not the whole mechanism.

    Evidence: 25 September 2026 · Türkiye's CDS tops 250 for the first time since 27 May: as the $18 billion liquidation of 131 funds proceeds, Erdoğan says no systemic risk; 45 arrested
  • Scenario

    Confirmed

    The published claim

    H3 · 20% — The demand diagnosis wins: the Fed moves to a harsher, front-loaded path, the dollar strengthens, and emerging-market premia widen indiscriminately.

    What happened

    On 23 September Barr and Collins signalled an additional hike; the probability of an October hike rose to 70% and the dollar index moved above 101. In Latin America premia kept widening: Argentina's country risk rose from 533 on 21 September to 609 basis points on 25 September. The decisive test of the data leg is the PCE release on 30 September.

    Evidence: 25 September 2026 · Argentina's country risk jumps 31 points in a day to 609, a 6-month high; around $795 million paid to the IMF, reserves fall to $48.2 billion
  • Scenario

    Still open

    The published claim

    H2 · 30% — The calendar passes cleanly: Argentina makes its 25 September payment and the review concludes favourably, the 28 September Washington round yields a tariff cut, Brazil's election ends with a market-friendly framework; the widening of 21 September is reversed.

    What happened

    Only one of the three triggers came due, and it did not produce the effect we expected: Argentina paid the IMF roughly $795 million, but on the same day country risk rose 31 points to 609 and reserves fell to $48.2 billion. The assumption that the payment would lower the premium proved wrong on this leg. Because the 28 September USMCA round and the 4 October Brazilian election have not yet taken place, the scenario remains open as a whole.

    Evidence: 25 September 2026 · Argentina's country risk jumps 31 points in a day to 609, a 6-month high; around $795 million paid to the IMF, reserves fall to $48.2 billion

What we learned

We got the outcome right and the ranking wrong: the report's main reading, repricing at bloc level, continued, but we gave the scenario that would carry it only 20%. There is also a mechanism error in the Argentina leg: we treated the payment calendar as the source of risk, yet the premium widened on the day the payment was made; the market priced not the payment itself but the reserves left after it. In the Türkiye leg, we did not foresee that a country-specific channel would enter the chain. From now on, in scenarios built on a payment calendar, the post-payment reserve level will be written as a separate watch indicator.

Published on 18 September 2026 · Reviewed on 26 September 2026

The buffer is thin in two places at once: European storage at 68.5% while the Asian spot climbs to 30 dollars

Open the report ›
  • Causal chain

    Confirmed

    The published claim

    Premise: European storage enters winter at 68.5% full, about 16 points below the seasonal average; the gap is a structural fragility.

    What happened

    The gap has not closed. On gas day 23 September EU storage held 796 TWh of 1,132 TWh capacity, 70.3% full; that is 15.7 points below the 2021–2025 average of 86.0% and also below the five-year low of 72.6%. Germany, at 57.12%, is 19.3 points behind the same day of 2025. In the six days since 17 September the fill level rose 1.8 points, while the gap to the average narrowed from about 16 points to 15.7 points, only a few tenths of a point.

    Evidence: 25 September 2026 · EU gas storage reached 70.3% on 23 September: 15.7 points below the five-year average and 11.7 below 2025; Germany stuck at 57%
  • Scenario

    Still open

    The published claim

    H1 · 50% — The two-way squeeze persists: the Hormuz constraint and reduced Qatari LNG flows continue, the EU storage gap stays in the 16-point band; Europe keeps competing with Asia for cargoes.

    What happened

    Both conditions of the trigger are in place as of 25 September: transits observed through Hormuz on 24 September were 9, with a ten-day average of 18; before the war there were about 125 large vessels a day. The EU gap stands at 15.7 points. Iran's proposal of 25 September to open within 7 days is only a proposal; the US did not treat it as a breakthrough. Because the scenario was defined over the whole winter, the verdict is open; we have not recorded data to measure the effect of the cargo competition on the Asian side.

    Evidence: 24 September 2026 · Tracked Hormuz transits fell to 9 on 24 September, 8 of them outbound; 23 September revised from 10 to 14, while The National gave 13 for the same day
  • Scenario

    Still open

    The published claim

    H2 · 25% — Supply partly normalises: Norwegian maintenance ends, Qatari flows through Hormuz partly reopen and spot LNG falls towards a 10-dollar floor.

    What happened

    The Hormuz condition did not materialise: according to Kpler data only 3 commodity vessels transited the strait on 22 September, and on 23 September the Cape Dao was hit and a seafarer was killed. We have written no records since publication on the end of Norwegian maintenance or on the Asian spot LNG price. The branch did not advance in this window, but it has not reached maturity.

    Evidence: 23 September 2026 · Bulk carrier Cape Dao hit by two projectiles in the Strait of Hormuz, 1 Indian seafarer killed; IMO counts 83 attacks and 23 deaths since the conflict began
  • Scenario

    Still open

    The published claim

    H3 · 25% — Cold snap and second outage: an early cold snap accelerates withdrawals while seven fields close in Libya or an Armenia-type outage recurs; the outage passes directly into price.

    What happened

    The Libyan leg was partly triggered: on the night of 21 September valve no. 7 on the line linking Sharara to Zawiya was closed and the NOC reported a loss of about 130,000 barrels a day; on 23 September the Zawiya refinery stopped. The closure of seven fields did not happen, and there is no cold snap. Brent rose to 106.60 dollars on 24 September, but the Houthi missile strike on Yanbu the same day makes it impossible to separate the price move from Libya.

    Evidence: 22 September 2026 · Valve number 7 on Libya's Sharara pipeline is shut and the NOC reports a loss of 130,000 barrels a day
  • Watched indicator

    Still open

    The published claim

    Threshold: EU storage falling below 65%, the zone where Europe would keep competing for cargoes out of necessity rather than choice.

    What happened

    The threshold was not approached; the level moved away from it: the fill rate rose from 68.5% on 17 September to 70.3% on 23 September. As the withdrawal season has not begun, the threshold has not yet been tested. Sources diverge: Global Energy Flow sees the daily pace of 0.21 points as falling short of the 80% target for 1 November, while Voltstack considers winter preparation on track at the current base.

    Evidence: 25 September 2026 · EU gas storage reached 70.3% on 23 September: 15.7 points below the five-year average and 11.7 below 2025; Germany stuck at 57%

What we learned

The European half holds: the 16-point gap narrowed only to 15.7 points in a week. But the report's headline was that the buffer is thin in two places, and we never measured the second place, Asian spot LNG, after publication; there is not a single record for the 30 dollars/mmBtu level, Pakistan's October–December cargo count or the BKMEA survey. As a result none of the three scenarios could be ruled on. We must also record the counter-reading of our own threshold: one source sees winter preparation on track at the current pace; by that source's measure, the thin-buffer thesis may be overstated. Method change: no report with a two-legged thesis will be published without at least one automatically or weekly tracked series defined for each leg; for two-channel triggers such as H3, the indicator that isolates the price effect is written down from the outset.

Published on 18 September 2026 · Reviewed on 26 September 2026

From a struck distillation unit to 100-lira diesel: Russian refinery capacity and Türkiye's three-channel bill

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 50% — The shortage becomes entrenched: repairs at AVT-3 and AVT-4 drag on, capacity does not return at the facilities hit, the ban is extended beyond 31 October; pump price rises in Türkiye continue at intervals.

    What happened

    The Russian side was more than confirmed: on 20 September two distillation units at the Moscow Refinery burned, on 21–22 September Kuibyshev and Ufa were hit, and Ukraine's General Staff announced that more than 45% of Russian refining capacity was out of action. The date of the ban leg is 31 October and has not yet arrived. The Türkiye leg, however, did not play out as we wrote: diesel fell from 100.40 lira on 17 September to 90.90 lira on 24 September and returned to 93.45 lira on 25 September after a 2.55-lira rise. The pump tracked Brent and the exchange rate, not Russian capacity.

    Evidence: 22 September 2026 · Ukraine strikes the Kuibyshev and Ufa refineries: more than 45% of Russian refining capacity is out of service
  • Scenario

    Falsified

    The published claim

    H2 · 30% — Gradual normalisation: repairs progress, the pace of attacks slows and the ban ends on 31 October without extension.

    What happened

    The opposite of the condition happened. Two days after publication, on 20 September, both distillation units at the Moscow Refinery, of 21,400 and 18,800 tonnes a day, burned and the facility stopped processing; Russia said it downed 1,110 drones the same night. On 22 September the capacity loss exceeded 45%, and on the night of 23 September Ufa was attacked again. The pace did not slow; it escalated.

    Evidence: 20 September 2026 · Both crude distillation units at the Moscow Refinery burn and the plant halts: Russia says it downed 1,110 drones overnight
  • Scenario

    Partly borne out

    The published claim

    H3 · 20% — Second shock wave: new deep strikes also cut the remaining primary distillation capacity, or the exchange rate passes 50 lira; the intervals between price rises shorten and competition for substitute supply hardens.

    What happened

    The first half of the trigger materialised within four days: primary distillation capacity at Moscow, Kuibyshev and Ufa went out of action on 20–22 September. The currency half did not; the dollar/lira rate set a record at 48.84 on 23 September but stayed below 50. The expected effect was not seen in Türkiye: the intervals between price rises did not shorten, and a 5.5-lira cut came on 24 September. The trigger to which we gave 20% materialised; the outcome we tied to it did not.

    Evidence: 20 September 2026 · Both crude distillation units at the Moscow Refinery burn and the plant halts: Russia says it downed 1,110 drones overnight
  • Causal chain

    Still open

    The published claim

    Step 2: Even if the dollar price of the product stays flat, the exchange rate raises the lira cost of imported fuel; the indicators are how close the dollar/lira rate is to the 50-lira threshold and successive diesel price rises.

    What happened

    The currency moved as in the report: the dollar/lira rate rose from 48.67 on 17 September to 48.82 on 21 September and 48.84 on 23 September, setting new records twice. The 50 threshold was not crossed. The series of successive price rises, however, was broken: the cut on 24 September was attributed to Brent falling to 98.53 dollars, and the rise on 25 September was explained by Brent's return to 100.79 dollars and an exchange rate of 48.87. The currency channel is working but has not reached the threshold.

    Evidence: 23 September 2026 · Dollar hits new high of 48.84 lira on 23 September; calculations show CBRT reserves fell $13.9 billion in 4 weeks to $174.5 billion
  • Watched indicator

    Still open

    The published claim

    Threshold: Türkiye's 10-year benchmark yield exceeding 34% would signal that domestic borrowing costs will create a second jump in the interest line of the budget deficit.

    What happened

    The threshold was not crossed: the 10-year yield stood at 32.64% on 24 September. Pressure concentrated at the front end of the curve; the 2-year yield rose from 36.43% on 23 September to 37.09% on 24 September, overtaking the 37% policy rate. The horizon was written as months; the verdict awaits monthly budget outturns.

    Evidence: 24 September 2026 · BIST 100 falls 2.74% to 12,888 points on 24 September as the 2-year lira bond yield climbs to 37.09%: Fed and oil pressure hit Turkish assets

What we learned

We read the loss of Russian capacity correctly, but its transmission to Türkiye at the wrong speed. The report's headline was 100-lira diesel; within a week the price fell to 90.90 lira, because in this window the pump tracked Brent and the exchange rate, not lost Russian cargoes. We wrote that the product premium would decouple from crude but never measured the Mediterranean diesel premium; we tied the chain's first link to a variable we did not measure. The second error is in scenario design: H1 and H3 shared a trigger, H3's trigger materialised in four days and two scenarios partly held at once. Henceforth: (1) a product-premium claim is written only with a premium series we track, (2) scenarios must be mutually exclusive at trigger level, (3) the pass-through of a Russian supply shock to Türkiye's pump prices enters no verdict unless separated from the Brent and exchange-rate effect.

Published on 15 September 2026 · Reviewed on 26 September 2026

Ukraine–Russia: as momentum changes hands on the front, the war shifts to energy infrastructure

Open the report ›
  • Falsifier

    Confirmed

    The published claim

    Condition that breaks the chain: if the proposal for a pause in reciprocal attacks on energy infrastructure is accepted, Russian export capacity recovers, Ukraine's grid gets relief and refined product prices fall back.

    What happened

    The condition did not materialise; the reading stands. On 22 September Zelensky discussed an energy ceasefire with Trump and said he was ready for any form of it; the Kremlin called the proposal a good idea but put forward sanctions and tanker security as preconditions. The same night Russia struck the Naftogaz facility in Poltava with at least 5 ballistic missiles. The energy ceasefire Trump floated on 14 September found no response on the ground.

    Evidence: 22 September 2026 · Zelensky discusses an energy truce with Trump; the same night Russia strikes a Naftogaz facility in Poltava with at least 5 ballistic missiles
  • Scenario

    Partly borne out

    The published claim

    H1 · 60% — The infrastructure war continues: the front stays largely static, both sides keep attacking energy infrastructure; Russian product exports stay low and Ukraine's grid enters winter fragile.

    What happened

    The infrastructure leg was confirmed in both directions. Ukraine hit the AVT-3 unit at YANOS on 17 September, the Moscow Refinery on 20 September, and Kuibyshev and Ufa on 21–22 September; the General Staff announced that more than 45% of Russian refining capacity was out of action. Russia struck the Naftogaz facility in Poltava on 22 September and data centres in Kyiv on 23 September. We published no territorial data to check the static-front leg; this half could not be audited.

    Evidence: 22 September 2026 · Ukraine strikes the Kuibyshev and Ufa refineries: more than 45% of Russian refining capacity is out of service
  • Scenario

    Still open

    The published claim

    H2 · 20% — Mutual pause: a reciprocal pause in attacks on energy infrastructure is accepted; Russian export capacity recovers and refined product prices fall back.

    What happened

    It did not happen within eleven days and the pace of attacks increased; on the night of 20 September Russia said it had downed 1,110 drones. But the branch is not closed: Moscow left a conditional door open, and on 23 September, after meeting Lavrov, Rubio said both sides had shown interest in a limited ceasefire covering grain and energy targets. As no horizon was written, no verdict was given.

    Evidence: 23 September 2026 · Russia's Krasnodar region declares a state of emergency over attacks on Black Sea ports: 90% of grain export capacity was knocked out in August
  • Scenario

    Still open

    The published claim

    H3 · 20% — The front gets moving: Ukraine's Lyman offensive widens or Russia launches a new major offensive; territorial change accelerates.

    What happened

    After 15 September we have no published event record on territorial change at Lyman or along the front line; nor do we yet have September territorial balance data. No verdict was given without evidence.

  • Causal chain

    Partly borne out

    The published claim

    Step 1: Port and refinery strikes constrain Russian export revenue more effectively than sanctions; the indicator is whether loadings at Novorossiysk and Ust-Luga recover.

    What happened

    A strong sign came that the port channel remains impaired: on 23 September the Krasnodar region declared a state of emergency backdated to 12 August, and August's attacks were reported to have knocked out about 90% of the region's grain export capacity. But the indicator we tracked was oil loadings; Novorossiysk's daily shipment figure could not be verified, and we recorded no data at all for Ust-Luga.

    Evidence: 23 September 2026 · Russia's Krasnodar region declares a state of emergency over attacks on Black Sea ports: 90% of grain export capacity was knocked out in August
  • Watched indicator

    Still open

    The published claim

    Threshold: EU gas storage remaining below 80%, the zone where Europe enters winter more exposed to attacks on energy infrastructure.

    What happened

    EU storage stood at 70.3% on gas day 23 September: 15.7 points below the 2021–2025 average and 11.7 points below 2025; Germany is at 57.12%. It remains below the threshold. The daily fill pace is about 0.21 points; sources diverge on whether the 80% target for 1 November will be reached. As the horizon is winter, the verdict is open.

    Evidence: 25 September 2026 · EU gas storage reached 70.3% on 23 September: 15.7 points below the five-year average and 11.7 below 2025; Germany stuck at 57%

What we learned

The main reading, that competition has shifted from territory to energy infrastructure, was more than confirmed within eleven days; the capacity loss exceeded 45% and both sides kept hitting energy targets. Our weakness was failing to measure what we promised to measure: the report's premise rested on ISW territorial data, yet afterwards we published not a single territorial record, so the front half of H1 and H3 could not be audited. We wrote Novorossiysk and Ust-Luga oil loadings as an indicator to watch, then never recorded that data. Method change: every series written as an indicator to watch in a report will be tracked as an event record in the overnight pipeline; if it cannot be tracked, it will not be put on the watch list.

Published on 16 September 2026 · Reviewed on 25 September 2026

Two straits, one lever: why the Red Sea route is no longer a safe harbour while Hormuz is closed

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Prolonged siege: the East–West pipeline stays closed for weeks, the Salalah corridor is not applied; the physical squeeze continues.

    What happened

    The pipeline leg proved wrong: the line, shut by a drone attack on 13 September, restarted at low flow on 22 September, 9 days later; full capacity was reported to take weeks. The physical squeeze leg held: only 10 commodity vessels transited Hormuz on 23 September, and reopening the strait remained dependent on Iran–US negotiations. No evidence was found that the Salalah corridor has been applied.

    Evidence: 22 September 2026 · Saudi Arabia restarted the East-West pipeline, shut by a drone attack on 13 September, at low flow on 22 September; full capacity will take weeks
  • Scenario

    Partly borne out

    The published claim

    H2 · 25% — Partial reopening: the pipeline reopens within days and the Iran–Oman corridor operates in a limited way; the spot price falls rapidly towards futures.

    What happened

    The pipeline leg materialised (22 September, low flow). The spot–futures gap narrowed but did not close: the FRED spot series fell from 130.80 dollars on 15 September to 114.89 dollars on 22 September, while futures stood at around 100 dollars on 22 September. Transits did not increase; there are no data showing the Iran–Oman corridor operating, and all vessels exiting on 22 September used the northern (Iranian) corridor. On 24 September the Houthis firing a ballistic missile at Yanbu made the export port of the reopened line a target as well.

    Evidence: 24 September 2026 · Houthis fire 6 ballistic missiles at Yanbu and Taif; with the Red Sea end of the East-West pipeline targeted, Brent closes up 3.41% at $106.60
  • Watched indicator

    Partly borne out

    The published claim

    Threshold: Brent exceeding 120 dollars would mark the zone where futures start to track spot and the disruption is no longer seen as temporary.

    What happened

    In the series in which we set the threshold (FRED spot), 120 dollars was exceeded after publication as well: 127.84 dollars on 16 September, 121.18 dollars on 17 September. But what we wrote as the meaning of the threshold, futures tracking spot, did not happen: Brent futures have not approached 120 dollars since publication, with the highest close at 106.60 dollars on 24 September. We defined the threshold in one series and its meaning in another.

    Evidence: FRED — Crude Oil Prices: Brent - Europe (DCOILBRENTEU)
  • Watched indicator

    Still open

    The published claim

    Threshold: daily transits through Hormuz exceeding 40 vessels would be the first physical evidence that the corridor is actually operating.

    What happened

    The threshold was not crossed; this is consistent with the siege reading that the corridor is not operating, but the threshold itself has not yet been tested: according to IMF PortWatch, the highest daily transit count between 16 and 20 September was 7; according to Kpler's revised data, 4 commodity vessels transited on 21 September, 7 on 22 September and 10 on 23 September. Kpler's initial figures are revised upwards by as much as double the following day.

    Evidence: 24 September 2026 · 10 commodity vessels crossed the Strait of Hormuz on 23 September, 9 of them inbound to the Gulf; preliminary data for 22 September revised from 3 to 7 ships
  • Watched indicator

    Still open

    The published claim

    Threshold: daily transits through Suez falling below 35 vessels would show territorial control on the Red Sea route feeding through to trade.

    What happened

    The threshold was not crossed after publication: according to IMF PortWatch, Suez transits on 16–20 September were 37, 37, 51, 43 and 42 vessels. 27 vessels transited Bab el-Mandeb on 23 September. The impact of territorial control on trade cannot yet be measured; the threshold remains under watch.

What we learned

Our biggest error was the duration of the pipeline outage: we wrote 'closed for weeks' in the most likely scenario, and the line reopened in 9 days. We correctly saw that the real source of the siege was the strait itself rather than the pipeline; transits came nowhere near 40 vessels. The second error was methodological: we defined the 120 dollar threshold in the FRED spot series but tied its meaning to the futures price, which is why the threshold appears both crossed and not crossed at the same time. From now on, every threshold will be written together with a single series and that series' meaning. The probabilities stand as written.

Published on 15 September 2026 · Reviewed on 25 September 2026

The technology front before the summit: a model distillation accusation, China's memory breakthrough and phase-two chip tariffs

Open the report ›
  • Scenario

    Partly borne out

    The published claim

    H1 · 55% — Truce extended, technology front hardens: the tariff truce is extended by a year at the summit, while chip controls and distillation measures remain.

    What happened

    The direction held, the duration did not. The truce was extended, but by two months rather than a year: on 23 September Bessent moved the end date from 10 November to 10 January. There was no easing of chip controls; the issue was not raised at the summit, and the hardening is coming not from the executive but from the 3 chip bills in the Senate defence package (committee votes 42-0, 36-8, 42-2). No statement on distillation measures could be found from the summit; this part could not be verified.

    Evidence: 24 September 2026 · No joint statement at Xi–Trump summit: tariff truce extended by just 2 months from 10 November to 10 January, no new decisions on chips or rare earths
  • Scenario

    Falsified

    The published claim

    H2 · 20% — Comprehensive détente: a tariff cut and permission for some chip sales are announced at the summit.

    What happened

    No joint statement came out of the summit and no tariff cut was announced; average tariffs remained at 36.5% on the US side and 31% on the Chinese side. No new chip sales permission was granted. Bessent leaving a bigger package open for January pushes this scenario to a later date, but it did not materialise at the summit in the form we wrote on 15 September.

    Evidence: 24 September 2026 · AI front stalls at summit: no signed text for an incident notification line, chip export controls not tabled, Congress pushes tightening with 3 bills
  • Scenario

    Falsified

    The published claim

    H3 · 25% — Rupture: the summit is cancelled or China responds to the distillation warning with a concrete countermeasure; the truce expires on 10 November.

    What happened

    The summit took place on 24 September, China announced no countermeasure and the truce will not expire on 10 November: the end date was postponed to 10 January. China's preparations to restrict 7 additional rare earth metals were also moved to 10 January 2027.

    Evidence: NBC News — US and China extend trade truce
  • Causal chain

    Confirmed

    The published claim

    Step 1: once model access becomes a security issue, expectations of easing in chip export controls weaken; the gauge to watch is whether the joint summit statement addresses artificial intelligence and chip controls.

    What happened

    No easing came: no joint statement was published, chip controls did not come to the table, and as of 24 September there was no signed text for the artificial intelligence incident notification line. H200 licences remained on paper, Chinese customs blocked purchases and Congress is preparing new restrictions.

    Evidence: 24 September 2026 · AI front stalls at summit: no signed text for an incident notification line, chip export controls not tabled, Congress pushes tightening with 3 bills
  • Watched indicator

    Still open

    The published claim

    Threshold: the dollar/lira exceeding 52 would mark the zone where global technology and trade tensions feed through to emerging market currencies.

    What happened

    The threshold was not crossed: the dollar/lira stood at around 48.87 on 24 September. The market reaction after the summit was limited; the S&P 500 fell 0.02% and bond yields set the tone of the day. As the truce has been extended to 10 January, this threshold will be checked again up to that date.

What we learned

We read correctly that the summit would take place and that the technology front would not soften; on the duration of the truce, however, we were too optimistic. While we expected a one-year extension, the parties chose a two-month one; this shows a preference for renewal at brief intervals over a lasting solution, and makes H1 only partly right. We also placed the source of the hardening in the wrong place: Congress, not the executive. Lesson: in truce scenarios, write the duration as a separate claim and state which institution the decision will come from. The probabilities of 20% for H2 and 25% for H3 stand as written; the next check date is 10 January 2027.

Published on 16 September 2026 · Reviewed on 24 September 2026

Three central banks tighten in the same week: answering a supply shock with rates, and Türkiye's carry flows

Open the report ›
  • Causal chain

    Confirmed

    The published claim

    Premise: the Fed tightened on 16 September and the ECB on 10 September; the Bank of Japan is expected to raise its rate to 1.25% on 18 September. Three major central banks are responding to an energy-driven supply shock in the same week by suppressing demand.

    What happened

    The expected third leg materialised: on 18 September the Bank of Japan raised its rate from 1% to 1.25%, the highest in 31 years, by a 7–2 vote. The premise that three central banks tightened within the same weekly window thus became fact.

    Evidence: Al Jazeera — Bank of Japan raises rates to 31-year high of 1.25%
  • Causal chain

    Confirmed

    The published claim

    Step 1: tightening does not stay confined to near-term rates; fiscal deficit concerns also pull the far end of the curve higher. The gauge to watch is the US 30-year yield relative to the 5.5% threshold.

    What happened

    Transmission was delayed at first, then arrived. On 17 September the 10-year yield eased from 5.04% to 4.94%, and the 30-year fell to 5.29% on 21 September; in that window the chain appeared not to hold. On 23 September the 70 billion dollar 5-year auction tailed at 5.033%, the 10-year yield rose to 5.10%, its highest since 2007, and the 30-year to 5.39%. The far end was pulled higher; the 5.5% threshold, however, has not yet been breached.

    Evidence: 23 September 2026 · US Treasury sells $70 billion of 5-year notes at 5.033%; after hot PMIs the 10-year yield hits 5.10%, its highest since 2007
  • Scenario

    Still open

    The published claim

    H1 · 55% — Patient tightness: the Fed delivers one more hike by year-end, the CBRT holds its rate, and oil stays above 100 dollars.

    What happened

    No verdict was issued as the horizon is year-end. The oil leg is oscillating around the line: Brent futures fell from 104.82 dollars on 17 September to below 100 dollars on 22 September (98.44–99.25 dollars depending on the source), then rebounded by about 4% to 103.31 dollars on 23 September.

    Evidence: 22 September 2026 · Brent falls from 104.82 dollars on 17 September to 100.05 dollars on 22 September as Hormuz transit data conflict
  • Watched indicator

    Still open

    The published claim

    Threshold: Türkiye's 5-year CDS exceeding 250 basis points would mark the zone where the risk premium fully prices in the energy bill channel.

    What happened

    The threshold was not breached, but the direction matches the report's reading: CDS stood at 232.97 basis points on publication day, 16 September, and rose to 242.43 on 21 September. Over the same period the dollar/lira stayed below the 52 threshold at 48.81 on 22 September.

    Evidence: 21 September 2026 · The dollar reaches an all-time high of 48.82 lira as the BIST 100 falls 1.92% at midday to 13,028.71 points

What we learned

We read the central bank calendar and the transmission to the far end of the curve correctly, but not the timing: in the first week after publication yields fell, and on 21 September we came close to marking this step as 'did not hold'. The verdict only became clear with the 23 September auction. Lesson: test a term premium claim against a named threshold and date, not within a one-week window. H1's 100-dollar condition broke for one day and returned the next; it will be tested again at the next review.

Published on 16 September 2026 · Reviewed on 23 September 2026

The buffer is gone, tightening has begun

Open the brief ›
  • Falsifier

    Confirmed

    The published claim

    Long-term borrowing costs are no longer set by central banks but by defence and coalition arithmetic. This reading is disproved if the French–German 10-year spread falls durably below 80 basis points, the UK 30-year gilt yield drops at least 50 basis points below 5.89%, and the Bundestag cuts Germany's 2027 defence line.

    What happened

    None of the three conditions occurred; the opposite did. On 19 September the French–German 10-year spread passed 100 basis points for the first time since 2012, and on 22 September it reached 102, putting Paris 12.7 basis points above Italy's 89.3. There was no easing in the UK either: August borrowing rose to £18.3bn and debt interest payments hit a record £8.8bn. There is no Bundestag decision cutting the defence line. The reading stands.

    Evidence: 19 September 2026 · France-Germany 10-year yield spread passes 100 basis points for the first time since 2012
  • Scenario

    Partly borne out

    The published claim

    S1 · 50% — Prolonged siege: Hormuz stays shut, the corridor is not applied, the East–West line stays closed for weeks; the Fed makes one more hike by year-end.

    What happened

    The siege leg held: on 21 September the 72,825 dwt tanker LR Stephanie was struck in Hormuz, weekend transits fell to 17 ships and visible commodity traffic to 2 ships on 21 September. The Fed leg is not settled: on 17 September the dot plot's 2026 median rose to 4.00–4.25%, leaving the door to a further hike open, but no hike was made. Half the scenario has occurred; the other half stays open until year-end.

    Evidence: 21 September 2026 · The 72,825 dwt tanker LR Stephanie is struck in Hormuz and two seafarers wounded; weekend transits fall to 17 ships
  • Scenario

    Still open

    The published claim

    S2 · 22% — Partial opening: the Iran–Oman corridor is applied, the blockade eases partly, the East–West line reopens within days.

    What happened

    The scenario is not dormant, but it did not happen. On 22 September Iran put on the table an offer to open Hormuz within seven days and the US and Iranian delegations met for three hours at the UN; the same day a parliamentary committee approved articles imposing a penalty of 20% of cargo value on ships in breach. There is an offer but no execution: the corridor was not applied and the blockade did not ease. The horizon has not passed.

    Evidence: 22 September 2026 · Iran puts an offer to reopen Hormuz within 7 days on the table; the same day a parliamentary committee approves fines of 20% of cargo for violators
  • Watched indicator

    Partly borne out

    The published claim

    To watch: the gap between physical spot Brent and the futures contract. Futures rising while the gap stays open would mean the market has stopped treating the disruption as temporary.

    What happened

    The indicator moved, but not in the expected direction. The futures contract fell from $104.82 on 17 September to $100.05 on 22 September, meaning the market kept treating the disruption as temporary; the rise in futures the test was looking for did not happen. The other end of the gap could not be measured: FRED's Brent spot series is stuck at 15 September because of publication lag. Half the indicator could be read, half could not.

    Evidence: 22 September 2026 · Brent falls from 104.82 dollars on 17 September to 100.05 dollars on 22 September as Hormuz transit data conflict

What we learned

A week of review yields two lessons. First: the core borrowing-cost reading held and the falsifier survived the test we set ourselves — which says less about the thesis being right than about the test doing its job. The second points at us: in writing the spread test we chose a two-legged indicator without allowing for one leg's source publishing days late. That is why, on 23 September, a visible marker for lagging series was added to the screen. From now on an indicator is written down together with its source's publication lag.

Published on 18 September 2026 · Reviewed on 23 September 2026

Scarcity is no longer distributed by price but by queue

Open the brief ›
  • Falsifier

    Confirmed

    The published claim

    Chip capacity is being placed by jurisdiction rather than by cost. This reading is disproved if second-stage chip tariffs are removed, export controls are mutually eased, or the Longtan expansion or the Dholera line is officially postponed.

    What happened

    No leg of the falsifier occurred; capacity went on being placed by jurisdiction. On 17 September Taiwan announced a 104-hectare expansion of the Longtan Science Park with NT$95.63bn of investment and 4,500 new jobs; the same day, at SEMICON India, a target of 50,000 wafers a month was announced for the Dholera line. No postponement. Nor was there mutual easing: on 20 September China's rare-earth magnet exports to the US were reported down 21% in August, to 512 tonnes. The reading stands.

    Evidence: 17 September 2026 · Taiwan is expanding Longtan Science Park by 104 hectares: NT$95.63 billion investment and 4,500 new jobs
  • Scenario

    Partly borne out

    The published claim

    S1 · 50% — Allocation pressure persists: the struck primary distillation units do not return within weeks and Russia's export ban is extended on 31 October.

    What happened

    The strike leg was confirmed and then some: with the Kuybyshev and Ufa refineries hit on 22 September, more than 45% of Russian refining capacity was out of action. The units did not return within weeks; the loss widened instead. The ban leg cannot yet be checked: the extension decision falls on 31 October and that day has not come. One leg of the scenario is settled, the other is open.

    Evidence: 22 September 2026 · Ukraine strikes the Kuibyshev and Ufa refineries: more than 45% of Russian refining capacity is out of service
  • Scenario

    Falsified

    The published claim

    S2 · 30% — Repair and easing: refinery repairs advance, the tempo of strikes falls, and the ban lapses on 31 October without extension.

    What happened

    The condition this scenario rested on did not occur; the opposite did. We expected repairs to advance and the tempo of strikes to fall; on 20 September both crude distillation units at the Moscow Refinery burned and the plant halted, and that same night Russia said it had downed 1,110 drones. By 22 September capacity loss passed 45%. In five days the tempo did not fall, it escalated; this branch is closed.

    Evidence: 20 September 2026 · Both crude distillation units at the Moscow Refinery burn and the plant halts: Russia says it downed 1,110 drones overnight
  • Watched indicator

    Still open

    The published claim

    To watch: the repair timetable for the AVT-3 unit struck at YANOS; AVT-4 has been under repair since 28 August, and how fast the two units return will determine when allocation pressure eases.

    What happened

    No checkable data has come in. There is still no announced repair timetable for AVT-3, and nothing has been said about AVT-4's return. Meanwhile the question has partly lost its meaning: what matters is no longer how fast the two units return but how fast new strikes land.

What we learned

Where we were wrong is clear. The 30% we gave S2 rested on the assumption that repairs would advance and the tempo of strikes would fall. Within five days the tempo did not fall but escalated. The error was not in the probability but in the choice of variable: we treated repair speed as primary when the decisive variable was the tempo of strikes. Making the repair timetable the watched indicator was the same mistake continued: on undefended infrastructure, repair time is secondary to strike frequency. In scenarios of this kind the primary indicator will from now on be the frequency of strikes and outages.