Global Rates
Brent crude
$84.90
-4.2% today, -10.0% in three sessions. Iran-Oman framework for a temporary Hormuz corridor and joint mine-clearing. The driver of everything else in this letter
ECB Sep 25bp hike
96.75c
+1.25 on a day cheaper energy should have softened it. At this price it is an execution trade, not an inflation trade
BoJ Sep 25bp hike
86.0c
-1.5, the largest move on the central-bank board. Cheaper crude removes much of the imported-inflation case. USD/JPY 158.97
US 10Y
4.639%
-6.5bp, Aug 25 settle, verified. 30Y 5.174%. Cuts our pos-017 fair value from 72 to 61 via the barrier
Bund / JGB / gilt
3.2550% / 2.8950% / 5.0676%
ALL THREE CARRIED from Aug 25 - our yield feed failed at source this morning and we are labelling that rather than estimating
Gold
$4,684.50
+1.0% while crude fell 4.2%. Two safe-haven assets disagreeing on the same headline is worth watching
One headline out of Tehran is repricing the global energy complex, and the two central-bank legs we track most closely are responding to it in opposite directions. Iran and Oman have agreed a framework for a temporary joint maritime corridor through the Strait of Hormuz plus a joint mine-clearing project, with Oman's foreign minister hoping to announce the interim corridor soon. Brent is $84.90, down 4.2% today and 10.0% from Friday's $94.39. The September ECB hike went UP on that news, from 95.5c to 96.75c. The September BoJ hike went DOWN, from 87.5c to 86.0c. Those two moves cannot both be about oil, and working out which one is not is the most interesting question on the board today. Our reading, offered as a reading and not a fact: cheaper crude is a terms-of-trade gift to Japan and removes a large part of the imported-inflation argument that has carried the BoJ's hiking case all year, so the BoJ leg falling is coherent. The euro area imports energy too, and by the same logic the ECB leg should have softened. It did the opposite, which suggests the September ECB hike is no longer an inflation trade at all but a done-deal trade at 96.75c, where the remaining 3.25 points are execution risk rather than economics. A caveat we have to publish rather than paper over: our sovereign yield feed failed at source this morning for Bunds, JGBs and gilts. Those three levels below are CARRIED from yesterday and labelled as carried. We are not going to infer this morning's Bund from this morning's oil price and present the result as a reading. The US 10-year, which we could verify, settled at 4.639%, down 6.5 basis points.
Today's Market Moves
ECB Interest Rates September 2026 - 25bp increase
95.5%→96.75%+1.25pp
Bid 96.7 / ask 96.8 on $92.6K, so this is tight and traded rather than a constructed mid. A euro-area hike becoming MORE likely on a 10% fall in imported energy is the anomaly of the day. The explanation we find most plausible is that at 96.75c the contract has stopped pricing the economics and is pricing whether a scheduled, well-telegraphed decision gets executed on Sep 10. There is nothing here for us: 3.25 points of upside against the entire stake is the wrong side of any Kelly calculation we run.
Bank of Japan September - 25bp increase
87.5%→86.0%-1.5pp
Bid 85 / ask 87 on $119K. The largest central-bank move on the board and the one that makes straightforward sense: Japan imports essentially all of its crude, and a 10% fall in the barrel takes a meaningful piece out of the imported-inflation argument that has driven the BoJ's normalisation case. USD/JPY at 158.97 cuts the other way and is why this is 86 and not 75. Watchlist only, no position, and we would want it below 70 before the risk-reward interested us.
Fed Decision in September - 25bp increase
33.5%→32.5%-1.0pp
Down a point on $15.5M of two-sided volume, with No change at 67.5c. This is the leg that moved sensibly on the oil news while the full-year Fed hike contract at 56.5c did not move at all - an internal inconsistency in our own book's favour that we are flagging rather than exploiting. CME FedWatch remains unverifiable at source for a sixth session; the last figure we could actually verify is 31.6% implied, dated Aug 20, and we carry it as stale.
10Y Touches 4.8% Before 2027 (pos-017)
67.0%→65.5%-1.5pp
The only global-rates market we hold. Bid 64 / ask 67. The 1.5 point fall is not the story; the eleven-point cut to our own fair value is, and it comes entirely from the 10-year settling at 4.639% rather than the 4.7050% intraday level we wrongly published yesterday. Barrier widens from 9.5bp to 16.1bp, model goes 72 to 61, and we are now marked above our own number.
10Y Touches 5.00% Before 2027 (watchlist)
17.5%→17.0%-0.5pp
Second consecutive session in which our consistency check fails. Model 31.1 against a 17.0c market, a 14.1pp gap, narrowed from 22pp. Fixed-volatility barrier models over-price far touches by construction and this is what that looks like. The standing rule holds: no fair-value increase on pos-017 until the check reproduces.
USD/JPY
159.14%→158.97%-0.17pp
Essentially flat, which is the quiet fact of the day. The joint intervention earlier this month, funded by selling euros, put a marker down near 160 and the pair has respected it through a session in which the BoJ hike probability fell a point and a half. A weaker yen normally argues FOR the BoJ moving; a lower oil price argues against. Today those cancelled and the hike leg fell anyway.
Screening Table
| # | Market | Expiry | Market Price | Fair Value | Gap (pp) | Direction | Volume | Confidence |
|---|---|---|---|---|---|---|---|---|
| 1 | 10Y Touches 4.8% Before 2027 (pos-017) | Dec 31 | 65.5% | 61% | -4.5pp | HOLD $25 YES - FV cut 72 -> 61, marked above our own number, new trim rule armed for below-4.60% settles | $$60K on the leg | 3/10 |
| 2 | BoJ September 25bp increase | Sep 18 | 86.0% | 84% | -2.0pp | NO POSITION - the oil move argues down and the yen argues up; not enough left at 86c either way | $$119K on the leg | 3/10 |
| 3 | Fed Decision in September - 25bp increase | Sep 16 | 32.5% | 31.6% | -0.9pp | NO POSITION - fair value 31.6 is STALE (Aug 20, FedWatch unverifiable at source for a 6th session); we do not trade off stale inputs | $$15.5M on the leg | 1/10 |
| 4 | ECB September 25bp increase | Sep 10 | 96.75% | 96% | -0.75pp | NO POSITION - 3.25 points of upside against the full stake; wrong side of any sizing rule we use | $$92.6K on the leg | 4/10 |
| 5 | 10Y Touches 5.00% Before 2027 | Dec 31 | 17.0% | 31.1% | +14.1pp | REFUSED for a 2nd session - the 14.1pp gap is real but the model producing it has failed its own consistency check twice; no stake | $$82.7K on the leg | 1/10 |
Top 5 Opportunities
1
10Y Treasury Touches 4.80% Before 2027 - the barrier widened by six and a half basis points and took eleven points of fair value with it — YES
↑ BUY YES-4.5pp
Market price
65.5%
Fair value
61%
Gap: -4.5pp
This is the only market on the global board where we have money, and today it went against us twice: once in price, from 67.0 to 65.5, and once in our own arithmetic, from a fair value of 72 to 61. The second is much larger and it is self-inflicted. Yesterday we recorded the US 10-year at 4.7050%, which was an intraday print, and computed a barrier of 9.5 basis points from it. The settle was 4.639%. The correct barrier is 16.1 basis points. In a barrier model that difference is nearly everything, and the position we described yesterday as five points cheap was in fact close to fair on the day we described it. What has not changed is the mechanism. Long-end pressure is a global condition, not a US one, and the three interventions that have capped it this month - the doubled Treasury buybacks, the joint yen intervention funded by selling euros, and the FIMA limit request - all have expiry dates or finite capacity. The buyback window runs out Nov 4 and the barrier runs to Dec 31. What has changed is that we no longer have an edge here by our own numbers, and we should say that in the same tone we used when the number was flattering.
▵ Bull case
- 16.1bp over 91 sessions still makes this roughly a two-in-three event on our own model, and the market at 65.5c broadly agrees.
- The buyback window that has capped the long end three times this month expires Nov 4, leaving around two months of the barrier period unprotected.
- Bunds near 3.2550% and gilts above 5% - carried figures - describe a global long end that is nowhere near comfortable.
- Today's yield fall is a second-order consequence of an oil headline, which is the least durable kind of yield move.
▿ Bear case
- We are marked 4.6pp above our own fair value, and the honest description of that is that we are holding something we think is expensive.
- The whole eleven-point cut traces to our own data error, which is a process failure and should reduce confidence in yesterday's letter more than in today's tape.
- If the Hormuz corridor is formally announced, more risk premium leaves the long end and the barrier widens further.
- The 5.00% leg check has now failed twice. The model is telling us we are rich today having told us we were cheap yesterday, and we do not fully trust either signal.
2
Bank of Japan September 25bp increase - the one central-bank leg that moved the way the oil price says it should — YES
↑ BUY YES-2.0pp
Market price
86.0%
Fair value
84%
Gap: -2.0pp
Down 1.5 to 86.0, bid 85 / ask 87, the biggest central-bank move on today's board and the only one with a clean explanation. Japan imports effectively all of its crude. A 10% fall in the barrel over three sessions is a direct improvement in the terms of trade and it removes a large piece of the imported-inflation argument that has underwritten the BoJ's normalisation path all year. Pulling the other way, USD/JPY at 158.97 is barely off the level that triggered the joint intervention earlier this month, and a weak yen is the other half of the same imported-inflation story. Those two cancelled today and the leg still fell, which tells us the market weighted the oil move more heavily. We have no position and are not taking one. At 86c the contract pays about 1.16-to-1 into a scheduled meeting three weeks away, and our fair value of 84 puts us on the wrong side of the market by two points. We keep it on the watchlist because it is the cleanest read available on whether the Hormuz de-escalation is being treated as durable, and because a fall below 70 would make the risk-reward genuinely interesting.
▵ Bull case
- USD/JPY at 158.97 remains near the level that provoked joint intervention, and yen weakness is the BoJ's most persistent argument for moving.
- JGB 10-year at 2.8950% - carried from yesterday - sits at a 1996 high, which is a market already positioned for normalisation.
- The meeting is scheduled for Sep 18 and the BoJ has telegraphed its direction repeatedly.
- A single three-session oil move is thin evidence on which to reverse a year-long policy trajectory.
▿ Bear case
- Crude down 10% is a direct terms-of-trade improvement for a country that imports all of it, and it cuts the imported-inflation case substantially.
- At 86c there are 14 points of upside against 86 points of downside on a three-week horizon.
- Our own fair value is 84, below the market, so we would be paying up to express agreement.
- If the Hormuz corridor is formally announced before Sep 18, this leg has further to fall.
3
ECB September 25bp increase - it went up on cheaper energy, which means it has stopped being an inflation trade — YES
↑ BUY YES-0.75pp
Market price
96.75%
Fair value
96%
Gap: -0.75pp
Up 1.25 to 96.75, bid 96.7 / ask 96.8. That is a tight, traded market, not a constructed mid, which makes the direction harder to dismiss. The euro area is a large net energy importer. A 10% fall in crude over three sessions is disinflationary for it in the most direct way available, and on the economics alone a September hike should have become slightly less likely today, not more. The reading we offer - and we want to be clear it is a reading, not something we have verified - is that at 96.75c this contract is no longer pricing whether the economics justify a hike. It is pricing whether a well-telegraphed decision gets executed on Sep 10. Once a contract crosses into that territory, incoming macro data mostly stops moving it, and the residual 3.25 points are procedural risk. That would explain why an inflation-relevant headline moved it the wrong way: it did not move it on inflation at all, it moved on the general de-escalation of tail risk. There is nothing here for us either way. We note it because a market that stops responding to its own fundamentals is information about the market, even when it is not a trade.
▵ Bull case
- Bid 96.7 / ask 96.8 on $92.6K is a tight two-sided market, so this price reflects real conviction rather than a thin mid.
- Bund 10-year near 3.2550% - carried - is consistent with a curve that has fully absorbed a September move.
- The decision is in fifteen days and has been telegraphed extensively.
- Lower energy costs support euro-area growth, which removes the growth objection to hiking even as it lowers headline inflation.
▿ Bear case
- 3.25 points of upside against the entire stake fails every sizing rule we use, at any confidence level.
- A contract that rises on disinflationary news is a contract whose price we cannot fully explain, and we do not put money into things we cannot explain.
- Our carried Bund level is a day old because our feed failed, so one of the inputs to this view is stale.
- Fifteen days is long enough for a formal Hormuz announcement and a full PCE cycle to change the picture.
4
Fed Decision in September 25bp increase - the leg moved sensibly while our own full-year contract did not, and we are flagging that rather than trading it — YES
↑ BUY YES-0.9pp
Market price
32.5%
Fair value
31.6%
Gap: -0.9pp
Down a point to 32.5c, with No change at 67.5c, on $15.5M of volume - by far the deepest market on this board. The interesting thing is the comparison with our own position. The September leg fell a point on the oil news. The full-year Fed Rate Hike in 2026 contract, which we own a runner in, did not move at all. Those two should be tied together: fewer September hikes with four months left mechanically reduces the full-year probability unless the market is simultaneously shifting hikes to October or December. It is possible that is exactly what happened. It is also possible one of the two has not repriced. We cannot distinguish between those from here, and we have flagged it in our US letter as an inconsistency that currently favours our book, which is the direction in which one has to be most careful. We are not taking a position on this leg for a much simpler reason: our only fair-value anchor is a CME FedWatch implied probability of 31.6% dated Aug 20, and we have now failed to re-verify FedWatch at source for six consecutive sessions. A six-day-old anchor against a market that has moved through a tariff escalation and an oil collapse is not an anchor. We publish it as stale and decline the trade for the seventh session running.
▵ Bull case
- $15.5M of two-sided volume makes this the most reliable price on the global board.
- Three FOMC members dissented in July in favour of hiking, so the September upside is real rather than theoretical.
- Core PCE at 3.29% YoY on the Cleveland nowcast is a long way from target with the meeting three weeks out.
- Today's July PCE release at 08:30 ET is a genuine catalyst that could move this leg in either direction before the meeting.
▿ Bear case
- Our fair value is a stale number and we know it. Trading off it would violate the rule we wrote about verification belonging to the same session as the flag.
- Crude down 10% reduces the near-term inflation impulse ahead of a meeting that is only three weeks away.
- Warsh speaks at Jackson Hole on Friday and could reset this entire leg in one sentence.
- The 0.9pp gap against a stale anchor is inside the noise of the anchor itself.
5
10Y Treasury Touches 5.00% Before 2027 - refused for a second day, and the reason is our model rather than the market — YES
↑ BUY YES+14.1pp
Market price
17.0%
Fair value
31.1%
Gap: +14.1pp
On today's numbers - 10-year at 4.639%, barrier 36.1bp, horizon sigma 38.16bp - the same barrier model that prices our own position returns 34.4% raw and 31.1 after the standard haircut, against a market of 17.0c. That is a 14.1pp gap, comfortably above our 10pp entry bar, and we are refusing it for the second consecutive session. The gap narrowed from 22pp yesterday to 14.1pp today, which is some comfort and not enough. Our objection is structural, not about the size of the number. Fixed-volatility barrier models over-price far touches by construction, because they assume the world stays as volatile at 5.00% as it is at 4.64%, when a sustained grind toward 5.00% would summon exactly the official response that stops it. We have watched three such responses this month. Our own 4.80% leg survives that objection because 16.1 basis points is inside ordinary weekly noise; a 36 basis point move to a round number that the Treasury has visibly defended is a different kind of bet. There is also a portfolio reason, and after today it is the stronger one. This book just watched one oil headline move four of its five positions in the same direction. Adding a fifth expression of the same long-end trade is not something we get to do on the day we published the invoice for the first four.
▵ Bull case
- The nominal gap is 14.1pp, above our stated entry bar, and it has now shown up on two consecutive sessions.
- Gilts above 5% - carried - prove a G7 sovereign can trade there, so 5.00% is not a theoretical level.
- The Treasury buyback window expires Nov 4 while the barrier runs to Dec 31, leaving a genuinely unprotected stretch.
- At 17c the payout is close to 5-to-1, so a small stake would need only a modest edge to be worth it.
▿ Bear case
- The model producing the gap has now failed its own consistency check on two consecutive days. We do not trade off a model we have publicly said we cannot reconcile.
- Fixed-volatility barrier models systematically over-price far touches, and 36bp to a defended round number is the textbook case.
- Three official interventions this month show the reaction function that would stop this move, and none of them are in the model.
- It would be a fifth correlated position in a book that today demonstrated exactly what correlation costs.