Global Rates
pos-017 fair value
61 (FROZEN)
Model says 66.8 on a 12.6bp barrier after the 10Y settled 4.674%. The Aug 25 rule blocks the increase because the 5.00% check now fails on TWO live legs by 17.6pp. Freeze costs 5.8 points.
German 10Y Bund
3.2648%
A 15-YEAR HIGH, and live again after two carried sessions. Ireland's 10-year hit a 12.5-year high the same morning. European long ends are leading this selloff, not following it.
JGB 10Y
2.9320%
Up 3.7bp from the 2.8950% we carried from Aug 25. Feed restored. BoJ September hike priced at 87.5c, up 1.5 on the week.
ECB September hike
97.6c
Up 0.85 from the 96.75 carried Wednesday. Effectively a done deal in the market's view, and now verified rather than inherited.
Fed September
30.5c hike / 68.5c hold
Live. The hike leg is down 2.0 from the 32.5 carried Wednesday, but sits close to the stale 31.6 FedWatch figure we have carried for eight sessions - two independent reads agreeing.
FedWatch
31.6 (STALE, 8th session)
Still unverifiable at source. Third-party quotes this morning conflicted outright, so we carry the dated figure rather than adopt a number we cannot check.
Both of this letter's data outages ended at once. The Polymarket feed, which was unreachable for the whole of Wednesday and left the entire central-bank board carried, is back; and the yield feed that had Bund, JGB and gilt carried from Aug 25 for two consecutive sessions is back too. So for the first time this week every number below is observed rather than inherited. That restoration immediately cost us an excuse. On Wednesday the 5.00% consistency check underneath pos-017 failed by 17 points, and we partly attributed the widening to an artefact: the 10-year had risen so our model lifted, while the market leg was carried and never re-observed. Today both legs are live, the 5.00% market prints 17.5c, the model says 35.1, and the gap is WIDER at 17.6 points with nothing carried to blame. The Aug 25 standing rule therefore binds harder, fair value stays frozen at 61 against an arithmetic 66.8, and the freeze now understates the position by 5.8 points instead of four. The global tape is doing what our position wants and we still would not raise the mark. The German 10-year hit a 15-year high at 3.2648%, Ireland's 10-year touched a 12.5-year high, the JGB is up 3.7bp to 2.9320% and the US 10-year settled 4.674%, provisionally, on Thursday. Long ends are selling off together, which is the environment in which a 12.6 basis point barrier gets touched. None of that changes the rule. Three central-bank legs also moved, all of them fresh: the ECB September hike is now 97.6c, the BoJ September hike 87.5c, and the Fed September hike 30.5c against 68.5c for no change. And a scheduling note that matters for how you read the rest: this letter now publishes at 04:06 ET, so Warsh's first Jackson Hole keynote and the BLS benchmark payroll revision, both at 10:00 ET, are previews here and not results.
Today's Market Moves
10Y Touches 5.00% Before 2027 (check leg)
17.0%→17.5%+0.5pp
THE DECISIVE NUMBER. Carried Wednesday, live today at 17.5c against a 35.1 model on the same barrier method we use for the 4.8% leg. 17.6pp apart on two fresh observations. Yesterday we could call the widening an artefact; today we cannot.
10Y Touches 4.8% Before 2027 (pos-017)
65.5%→64.5%-1.0pp
LIVE. The model moved the other way - 4.674% narrows the barrier to 12.6bp and lifts the arithmetic to 66.8 - while the market drifted a point lower. FV stays 61 by rule.
ECB September 25bp increase
96.75%→97.6%+0.85pp
Verified live. With the Bund at a 15-year high the September hike is close to fully priced, and the ECB leg has now spent a fortnight behaving as a rates trade rather than an inflation trade.
BoJ September 25bp increase
86.0%→87.5%+1.5pp
Live, and moving with the JGB, which is up 3.7bp to 2.9320% with USDJPY at 159.50. No change prices 12.5c and a 50bp move 0.95c.
Fed September 25bp increase
32.5%→30.5%-2.0pp
Live. Down 2.0 on the carried figure but within a point of the stale 31.6 FedWatch reading, which is mild reassurance that our carried Fed pricing has not drifted far from reality during the outage.
Zero Fed Rate Cuts in 2026 (pos-013)
86.45%→88.7%+2.25pp
Best mark since entry. Claims 203k against ~208k expected, Cleveland core nowcast holding 3.40%, Brent settling up 0.77%. FV 88 -> 89, with the trim armed at 90c now 1.3c away and this letter published before the US session.
Screening Table
| # | Market | Expiry | Market Price | Fair Value | Gap (pp) | Direction | Volume | Confidence |
|---|---|---|---|---|---|---|---|---|
| 1 | 10Y Touches 4.8% Before 2027 (pos-017) | Dec 31 | 64.5% | 61% | -3.5pp | HOLD $25 YES - model says 66.8, our rule freezes it at 61; the check failed again with nothing carried | $$60K on the leg | 3/10 |
| 2 | Zero Fed Rate Cuts in 2026 (pos-013) | Dec 31 | 88.7% | 89% | +0.3pp | HOLD $82 YES - FV 88 -> 89; trim armed 1.3c away and may fire unobserved | $$7.6M | 4/10 |
| 3 | ECB September 25bp increase | Sep 10 | 97.6% | 97% | -0.6pp | NO POSITION - fully priced at 97.6c; no edge worth a ticket | $$262K | 4/10 |
| 4 | BoJ September 25bp increase | Sep 18 | 87.5% | 86% | -1.5pp | NO POSITION - watchlist only; marked slightly rich and correlated with the rest of the book | $$355K | 3/10 |
| 5 | Fed September 25bp increase | Sep 16 | 30.5% | 31% | +0.5pp | NO POSITION - within noise of a stale FedWatch 31.6; we hold the 2026 contracts instead | $$55.7M on the event | 2/10 |
Top 5 Opportunities
1
10Y Treasury Touches 4.80% Before 2027 - long ends sold off worldwide and we still froze the mark — YES
↑ BUY YES-3.5pp
Market price
64.5%
Fair value
61%
Gap: -3.5pp
This is the day the global tape argued hardest for raising this fair value and we did not. The Bund printed a 15-year high at 3.2648%, Ireland's 10-year hit a 12.5-year high, the JGB rose 3.7bp to 2.9320% and the US 10-year settled 4.674% on Thursday, its second consecutive rise. A synchronised long-end selloff is precisely the environment in which a 12.6 basis point barrier gets touched, and the arithmetic agrees: 73.8% raw, 66.8 after the buyback haircut. But the Aug 25 standing rule says no fair-value increase until the 5.00% consistency check reproduces, and today the check failed in its cleanest form yet. Both legs are live for the first time since Tuesday. The 5.00% market prints 17.5c. The same barrier method that produces our tradeable 66.8 produces 35.1 there. The gap is 17.6 points and it is wider than yesterday's, which we had partly excused as an artefact of a carried market leg. That excuse is gone, and what remains is the plain reading: a fixed-volatility barrier model over-prices far touches by construction, so its output on our own leg is probably also too high, just by less. Fair value stays 61. Note what that costs. We are publishing a book that understates this position by 5.8 points on our own arithmetic, on a day when every external input moved in its favour.
▵ Bull case
- Bund at a 15-year high, Irish 10-year at a 12.5-year high, JGB +3.7bp - long ends are selling off together, which is how a 12.6bp barrier gets touched.
- The US 10-year has risen two consecutive sessions, 4.664% then 4.674%, moving toward the barrier.
- Model gives 66.8 after haircut against a 64.5 market, so even the frozen version of us thinks the market is not obviously expensive.
- The buyback window expires Nov 4 while the barrier runs to Dec 31, leaving nearly two months of unsupported duration.
▿ Bear case
- The consistency check now fails by 17.6pp on two fresh legs - the strongest evidence yet that the model over-prices touches.
- The 4.674% Aug 27 settle is provisional; FRED and H.15 still end at Aug 26 and publish today, so the arithmetic rests on a vendor number.
- Our house session count of 89 disagrees with a direct calendar count of 84, a 0.7 point difference and an unexplained error in our own method.
- European long-end weakness is being driven by supply and fiscal news rather than by anything that mechanically pulls US yields to 4.80%.
2
Bund at a 15-year high - the European long end is now leading the global selloff — n/a
↑ BUY YES0pp
Market price
0%
Fair value
0%
Gap: 0pp
The German 10-year at 3.2648% is a 15-year high, and it arrived on the same morning Ireland's 10-year touched a 12.5-year high and the gilt sat at 5.0491%. For two sessions we could not see any of these because the yield feed failed at source and we carried Aug 25 values with a label saying so; today they are observed again, and the two-session move is Bund +1.0bp from 3.2550%, JGB +3.7bp from 2.8950% and gilt -1.9bp from 5.0676%. The composition matters more than the levels. This is not a US-led selloff being imported into Europe. The ECB September hike is priced at 97.6c, European equity indices are higher this morning with the DAX up 0.69% and the CAC up 1.13%, and the Bund is making 15-year highs anyway, which points at supply, fiscal issuance and a genuinely hawkish ECB rather than at spillover. For our book the relevance is indirect but real: a world in which every developed long end is repricing higher is one in which the US 10-year is more likely to touch 4.80% before year-end, and that is the entire pos-017 thesis. We hold no European rates position and are not opening one - the book is already concentrated in hawkish-central-bank risk and adding a correlated European leg would deepen exactly the exposure we keep warning about.
▵ Bull case
- A synchronised developed-market long-end selloff raises the probability of the US 10-year touching 4.80%, which is the pos-017 thesis.
- The move is European-led rather than imported, which makes it less likely to reverse on a single US data print.
- ECB September priced at 97.6c gives a clear near-term catalyst that is already close to fully discounted.
- The yield feed is live again after two carried sessions, so this is observed rather than inferred.
▿ Bear case
- Fiscal and supply-driven European yield moves do not mechanically pull the US 10-year higher; the correlation is loose and episodic.
- We hold no European rates exposure, so this is commentary rather than a position, and we should not let it flatter the pos-017 thesis.
- A 15-year high is a level, not a direction; European long ends have reversed sharply from extremes twice this year.
- Adding any European leg would concentrate the book further into hawkish-central-bank risk that is already its dominant exposure.
3
ECB September 25bp increase - fully priced, which is itself the information — YES
↑ BUY YES-0.6pp
Market price
97.6%
Fair value
97%
Gap: -0.6pp
97.6c, up 0.85 from the 96.75 we carried on Wednesday and verified live today. There is no trade here and there has not been one for a fortnight - a contract at 97.6c pays 2.5 cents on the dollar for taking central-bank event risk, and our fair value of 97 is if anything below the market. What is worth recording is the consistency. On Tuesday we flagged that the ECB leg had stopped behaving like an inflation trade and started behaving like a rates trade: it rose on disinflationary news, which only makes sense if the market thinks the September move is locked regardless of the next print. Today the Bund makes a 15-year high, European equities rally, and the ECB leg goes up again. That is a market treating September 10 as settled and trading the path beyond it. For our book the read-across is to pos-010 and pos-013: a G3 central bank being priced at 97.6c to hike while the Fed's own September hike sits at 30.5c is a reminder that the hawkish repricing we are positioned for is already complete somewhere else, and incomplete in the one place we own it.
▵ Bull case
- 97.6c with the Bund at a 15-year high is internally coherent; the market is not confused about direction.
- The leg has risen through disinflationary news, which indicates conviction about September rather than sensitivity to the next print.
- Verified live today after being carried through Wednesday's outage, so the level is real.
- It provides a clean external check on our hawkish-central-bank thesis without our having to pay for it.
▿ Bear case
- At 97.6c the payoff is 2.5 cents for full event risk - no rational size fits.
- Our FV of 97 is below the market, so if we had a view it would be the wrong way round for a buyer.
- Volume of $262K is thin enough that the last print is not a reliable clearing level.
- A hawkish ECB is already fully discounted, so it offers no further support to the US legs we actually hold.
4
BoJ September 25bp increase - the JGB moved and the leg followed — YES
↑ BUY YES-1.5pp
Market price
87.5%
Fair value
86%
Gap: -1.5pp
87.5c for a September hike, up 1.5 from the 86.0 carried on Wednesday, with no change at 12.5c and a 50bp-or-more move at 0.95c. The JGB 10-year is 2.9320%, up 3.7 basis points from the 2.8950% we had been carrying since Aug 25, and USDJPY sits at 159.50. Those three facts are consistent with each other, which is the main thing worth saying after two sessions in which we could verify none of them. We mark fair value at 86 and therefore see the leg as mildly rich, but not by enough to trade against, and we would not take the other side of a hiking BoJ with the yen near 160 in any case. The reason this stays on the watchlist rather than becoming a position is the correlation cap. This book already owns a Fed hike, a no-cuts contract, a one-hike rate bucket and a 10-year barrier - four expressions of the same hawkish-rates idea. A BoJ hike leg would be a fifth. The fact that it looks slightly mispriced is not a reason to add it; the whole point of the cap is that it binds on the days when the correlated thing looks attractive.
▵ Bull case
- JGB up 3.7bp to 2.9320% with USDJPY at 159.50 - the yen and the curve both argue for a move.
- Verified live after two sessions carried, and the price is coherent with the JGB level.
- The 50bp tail at under 1c suggests the market sees a measured hike rather than a policy shock.
- A hiking BoJ supports the global long-end selloff that pos-017 needs.
▿ Bear case
- At 87.5c against our FV of 86 the leg is mildly rich, so there is no long edge.
- It would be a fifth correlated expression of hawkish-rates risk in a five-position book - exactly what the correlation cap exists to stop.
- $355K of volume is thin for a G3 central bank contract.
- The BoJ has repeatedly declined to move when the market priced it above 85c, and we have no informational advantage on that question.
5
Fed September - the live market and our stale FedWatch finally agree — YES
↑ BUY YES+0.5pp
Market price
30.5%
Fair value
31%
Gap: +0.5pp
The September board is live for the first time in two days: 68.5c no change, 30.5c a 25 basis point increase, 1.35c a cut. The hike leg is down 2.0 from the 32.5 we carried. The useful thing is the cross-check. We have been carrying a FedWatch reading of 31.6 as explicitly stale and dated for eight consecutive sessions, because the source remains unverifiable and third-party quotes this morning conflicted with each other outright - one summary reporting roughly 60% for a hike, another 68.4% for no change as of Aug 20, another 13.7% for a hike at the September 17 meeting. We are not going to adopt a number we cannot check, so 31.6 stays carried and labelled. But Polymarket's independent 30.5c lands within a point of it, which is the first evidence in over a week that our stale figure has not drifted far from reality. That is worth something and it is not worth much: two prices agreeing does not make either verified, and we will keep flagging FedWatch as stale until we can read it at source. No position - we express this view through the 2026 contracts, where the horizon is long enough that a single meeting does not settle it.
▵ Bull case
- Live at 30.5c and within a point of our carried FedWatch 31.6, suggesting the stale figure has not drifted.
- $55.7M of volume on the event makes this the most liquid Fed pricing we can observe directly.
- Coherent with the 2026 ladder, where the market puts roughly 48 points on some tightening by year-end.
- A cut priced at 1.35c supports pos-013, our no-cuts position, from an independent market.
▿ Bear case
- FedWatch remains unverifiable at source for an eighth session, so one of our two cross-checks on Fed pricing is still dated.
- Two prices agreeing is not verification; both could be wrong in the same direction.
- Warsh speaks today at 10:00 ET and could reprice the entire board hours after we publish.
- We hold no September position, so this is a reference point rather than actionable, and confidence is correspondingly low.