● Live
Daily US Global Rates Portfolio Archive Method

Global Rates

ECB Sep hike
94.5c
Up from 92c Friday, verified at source. Sep 10 decision; a 25bp increase is now close to fully priced
BoJ Sep hike
87.5c
Up from 84.5c, verified at source. Our completed miss: flagged at 40c on Aug 3, refused at 42.5c on Aug 10
Fed Sep hike
32.5c
+5.0 on the US-Canada tariff restart; no change at 66.5c. FedWatch unverifiable at source for a third session
10Y Bund
3.2502%
Near the highest since 2011. Gilt 5.0527%. Long-end repricing is not a US-specific story
10Y JGB
2.90%
+1.9bp overnight, back at levels last seen in 1996 as the September BoJ hike is priced at 87.5c
10Y US Treasury
4.70%
-3.8bp with Brent at $92.41 (-2.1%) as Bessent unveils the Iran sanctions package. 30Y 5.229%
The three-way divergence we described on Friday narrowed today, and it narrowed from the American end. Polymarket now prices a 25bp ECB increase on September 10 at 94.5c and a 25bp Bank of Japan increase on September 18 at 87.5c, both verified at source in this session and both higher than Friday's 92c and 84.5c. The Fed's September meeting is priced at 32.5c for a hike against 27.5c on Friday, with no change at 66.5c. Three developed-market central banks, one oil shock, and the gap between the two that are certainly moving and the one that might has closed by five points in a weekend. What closed it was not oil and not a central banker. It was trade. US-Canada negotiations collapsed late Friday night, with Washington imposing 50% tariffs on roughly $20bn of Canadian goods under Section 338 and Prime Minister Carney committing to dollar-for-dollar retaliation from September 8 across steel, dairy, appliances, agricultural equipment, pulp, paper and electronics. Carney's account is that the United States introduced late demands touching Canada's other trading relationships, its auto sector and protections for Canadian culture and the French language. For a rates book the mechanism is straightforward: a North American trade war is an inflation impulse, and the market added six points to Fed-hike-in-2026, taking it to 55.5c. Long ends did not confirm it. The US 10-year fell 3.8bp to 4.70% and the 30-year 4.7bp to 5.229%, both tracking crude rather than tariffs, because Treasury Secretary Bessent spent the day unveiling what he pre-announced as an economic D-Day sanctions package against Iran and the oil market read sanctions as the substitute for escalation rather than the prelude to it. Brent is $92.41, down 2.1%; WTI is $84.80, down 2.6%. Iran has responded by warning of ship seizures in the Strait of Hormuz and the rial is at a record low, so the risk is not gone, it is merely not today's price. Elsewhere the repricing continues without needing American help. The German 10-year Bund is 3.2502%, up fractionally and still near its highest since 2011. The 10-year JGB is 2.90%, up 1.9bp overnight and back in territory it last occupied in 1996, lifted by exactly the same September BoJ hike now priced at 87.5c. The 10-year gilt is 5.0527%, slightly lower on the day but above 5%. That is four sovereign long ends at or near multi-year and in one case multi-decade highs, which is the structural case for our only global-facing position and the reason a US buyback programme that expires on November 4 was never going to settle it. On pos-017, our bet that the US 10-year touches 4.80% before 2027, fair value moves from 70 to 72. We want to be precise about the character of that move because the last two sessions have not been: the yield fell, the barrier is now 10.0bp instead of 10.9bp, 93 sessions remain at about 4bp of daily vol for sigma of 38.6bp, and the same formula with the same haircut returns 72. It is arithmetic. After a 74-to-60 cut and a 60-to-70 reversal in consecutive sessions, a two-point mechanical adjustment is the whole of the improvement we are claiming. The market is 67 mid, 64 bid / 70 ask, the position is +$76.52 on a $25 stake, and the take-profit gate stays where we lowered it, at a 66c bid, two points away. The BoJ September hike we flagged at 40c on August 3 and declined to chase at 42.5c on August 10 now trades at 87.5c. It stays retired as a completed miss and the number keeps getting worse in public, which is the point of publishing it. No trades today. Book: 5 open, $332 staked, +$100.63 unrealized, +$611.08 realized, 11/14.
Today's Market Moves
ECB Sep 25bp increase
92.0%94.5%+2.5pp
Verified at source today. No change now trades at 3.55c and a 25bp cut is at 0.05c, so the September 10 decision is effectively settled in the market's mind. Germany is selling 30-year paper at the highest yields since 2011 and the Bund sits at 3.2502%. There is no position here and there is no edge at 94.5c — this is context for the term-premium mechanism behind pos-017, not a screen.
BoJ Sep 25bp increase
84.5%87.5%+3.0pp
Verified at source today, up three points. This remains the book's cleanest completed miss: we flagged it at 40c on Aug 3, declined to chase at 42.5c on Aug 10, and it is now 87.5c. Retired as a miss and reported every session it moves further away, because a miss that stops being mentioned once it hurts is not transparency. The 10-year JGB at 2.90% is the same trade expressed in the cash market.
Fed Sep 25bp increase
27.5%32.5%+5.0pp
Five points higher on the US-Canada tariff collapse, which is the only new inflation information in the weekend. No change is 66.5c. Refused for a fifth consecutive session, and the binding reason is unchanged: we could not verify a CME FedWatch reading at source for a third straight session because the tool is behind a registration wall, so the 36.6% secondary figure published Friday is not a tradeable input under our own Aug 6 rule. The notional 4.1pp gap is under the bar anyway, and the correlation cap blocks a fifth hawkish-Fed ticket independently.
10Y Touches 4.8% (pos-017)
66.5%67.0%+0.5pp
FV 70 -> 72, mechanically. US 10-year 4.70% (-3.8bp) leaves the barrier at 10.0bp; 93 sessions to Dec 31 at ~4bp daily vol gives sigma 38.6bp and 2*(1-Phi(0.2592)) = 79.6% raw; same Sep 9-Nov 4 buyback haircut as yesterday -> 72. Market 67 mid, 64 bid / 70 ask, +$76.52 on $25. Take-profit gate unchanged at a 66c bid, now two points away. The global corroboration is the actual thesis: Bund near a 2011 high, JGB at a 1996 high, gilt above 5%.
US-Canada tariffs (new context)
0%50%+50pp
New to the letter. Talks collapsed late Friday; 50% duties on about $20bn of Canadian goods under Section 338, with energy, potash and critical minerals carved out, and Canadian retaliation from Sep 8 on steel, dairy, appliances, agricultural equipment, pulp, paper and electronics. We treat this as a real but small inflation impulse — roughly $10bn of duties in a $30tn economy, with the shortest-pass-through categories deliberately excluded — which is why our Fed-hike fair value did not move even though the market's did.
Screening Table
# Market Expiry Market Price Fair Value Gap (pp) Direction Volume Confidence
110Y Touches 4.8% Before 2027Dec 3167.0%72%+5.0ppHOLD $25 YES — FV 70 -> 72 from the formula; under the bar, no add$$60K on the leg
6/10
2Fed Sep 25bp IncreaseSep 1632.5%36.6%+4.1ppNO TRADE — FV unverifiable at source for a third session; under bar; correlation-capped$$9.6M on the leg
3/10
3BoJ Sep 25bp IncreaseSep 1887.5%88%+0.5ppNO TRADE — retired as a completed miss; no edge at 87.5c$$118K on the leg
4/10
4ECB Sep 25bp IncreaseSep 1094.5%95%+0.5ppNO TRADE — effectively settled; context only$$92K on the leg
4/10
510Y Touches 5.0% Before 2027Dec 3116.0%16%0ppNO TRADE — internally consistent with the 4.8% leg at our own vol assumption$$83K on the leg
4/10
Top 5 Opportunities
1
US 10Y Touches 4.80% Before 2027 — the arithmetic moves two points and nothing else does — YES
Dec 31, 2026·$60K on the leg·Confidence ★★★☆☆ 6/10
↑ BUY YES+5.0pp
Market price
67.0%
Fair value
72%
Gap: +5.0pp
The only position in this book with a global thesis, and the one we have handled worst this week. On Wednesday we cut fair value from 74 to 60 after the US Treasury announced it would at least double long-end liquidity buybacks to $4bn per operation for the September 9 to November 4 window and the 10-year fell 6bp. On Thursday the market erased that entire move and on Friday we reversed the cut to 70, calling two double-digit revisions in two days instability rather than agility. Today fair value goes to 72 and we are writing that number down because a formula produced it, not because a story did. The US 10-year fell 3.8bp to 4.70%, following crude lower as Bessent unveiled the Iran sanctions package. The barrier is therefore 10.0bp rather than 10.9bp. Ninety-three business sessions remain to December 31. At roughly 4bp of daily volatility sigma is 38.6bp, and 2*(1-Phi(10.0/38.6)) gives a raw two-sided touch probability of 79.6%. Apply the same haircut we applied on Friday for the buyback window and fair value is 72. Same inputs, same haircut, one moved variable, two points of change. The global evidence is why we are relaxed about a US intervention that expires before the barrier does. The German 10-year Bund is 3.2502% and near its highest since 2011, with Berlin issuing 30-year paper at yields last seen that same year. The 10-year JGB is 2.90%, up 1.9bp overnight, at levels last seen in 1996, dragged there by a September BoJ hike the market now prices at 87.5c. The 10-year gilt is above 5%. Four developed sovereign long ends are repricing on deficits, issuance and inflation at the same time, and a secondary-market liquidity operation in one of them does not retire net supply in any of them. The market is 67 mid on a 64 bid and a 70 ask; the spread is seven points and the mid is a convenience. The position is +$76.52 on a $25 stake. The gate is unchanged and stated in advance for the fourth session running: take-profit at a 66c BID, currently 64c, and we will not raise it back to 72c simply because fair value recovered.
▵ Bull case
  • Barrier 10.0bp with 93 sessions left; raw two-sided touch probability 79.6%
  • Bund 3.2502% near a 2011 high, JGB 2.90% at a 1996 high, gilt above 5%
  • Buyback window ends Nov 4; the barrier runs to Dec 31, so December is unprotected
  • A North American trade war adds an inflation channel that argues for term premium
▿ Bear case
  • 64 bid / 70 ask on $60K — the 67 mid is not a level we could exit at
  • Oil fell 2.1% today and crude is the fastest route to 15bp lower in the 10-year
  • Third fair-value figure in three sessions, however mechanically today's was derived
  • Iran sanctions may yet end the war; a peace print takes yields down, not up
2
Fed September Hike — the divergence closes from the American end — YES
Sep 16, 2026·$9.6M on the leg·Confidence ★★☆☆☆ 3/10
↑ BUY YES+4.1pp
Market price
32.5%
Fair value
36.6%
Gap: +4.1pp
Friday's global letter was built around a divergence: the ECB at 92c and the BoJ at 84.5c against a Fed at 27.5c, three central banks facing one oil shock and only two of them moving. Today the gap closed five points from the American side, and the cause came from trade rather than from energy. US-Canada talks collapsed late Friday night, 50% tariffs now apply to roughly $20bn of Canadian goods, and Ottawa retaliates dollar-for-dollar on September 8. Fed-hike-in-2026 went from 49.5c to 55.5c and the September meeting from 27.5c to 32.5c. Whether that is the right size of move is a separate question and we think it is not: the tariff carves out energy, potash and critical minerals, and $10bn of duties in a $30tn economy is small next to Brent falling 2.1% on the same day. But we are not trading either side of it, and the reason we are not is procedural rather than analytical. For the third consecutive session we could not verify a CME FedWatch September probability at source. The tool sits behind a registration wall and the data endpoint errored. The 36.6% we quote is from a secondary source published on Friday. Under the rule we published on August 6, an unverified price is not a tradeable input, and that alone ends the discussion before the 10pp bar (the notional gap is 4.1pp) or the correlation cap (this would be a fifth position paying off on a hawkish Fed) are reached. We are close to removing this fair value from the letter entirely rather than reprinting an apology beside it each day, and we will say so when we do. One structural note for readers checking our numbers: hike-by-September at 33.5c reconciles with the 32.5c September meeting leg, but hike-by-October is unchanged at 40.5c against a full year at 55.5c, which compresses the September-to-October step and widens November-December implausibly. On a leg under $500K of volume we read that as a stale quote, not a view, and we are not building anything on it.
▵ Bull case
  • $9.6M of volume on the leg — genuinely tradeable size if the gap were real and verified
  • Warsh's first Jackson Hole keynote as Chair on Friday is a live repricing event
  • The tariff restart is a real inflation channel the September meeting must price
▿ Bear case
  • Fair value unverifiable at source for a third session — not tradeable under our own rule
  • 4.1pp is under the 10pp bar even at face value
  • Correlation cap: a fifth hawkish-Fed ticket in a five-position book
  • Brent fell 2.1% today; the dominant inflation input moved the other way
3
Bank of Japan September Hike — the miss keeps getting more expensive — YES
Sep 18, 2026·$118K on the leg·Confidence ★★☆☆☆ 4/10
↑ BUY YES+0.5pp
Market price
87.5%
Fair value
88%
Gap: +0.5pp
We flagged this market at 40c on August 3 as the clearest mispricing in global rates and did not buy it. We looked again on August 10 at 42.5c and declined to chase. On Friday we corrected a carried mark of 72c to a verified 84.5c. Today it is 87.5c, verified at source, with no change at 12.5c. That is a completed miss of roughly 47 points on a market we identified correctly and did not own, and we report it every session it moves further away because a miss that quietly stops being mentioned once it starts hurting is not transparency, it is editing. The mechanism was never obscure. The 10-year JGB is 2.90% today, up 1.9bp overnight and at levels last seen in 1996, and it has been climbing all month precisely because a September hike was becoming consensus. We had the thesis and the cash-market confirmation and we did not convert either into a position. There is no trade here now: at 87.5c against our fair value of 88 the gap is half a point, and the leg turns over about $118K, so even if there were an edge it would not survive the spread. It stays retired. The lesson we keep restating, because restating it is the only mechanism we have: our 10pp bar protects us from bad trades and it also, on a small number of occasions, protects us out of good ones. We keep the bar. We publish the bill.
▵ Bull case
  • 10-year JGB at 2.90%, a 1996 high, is the cash-market confirmation of the same view
  • No change trades at only 12.5c — the market treats the Sep 18 hike as near-settled
  • Global tightening bias is intact with the ECB at 94.5c for September
▿ Bear case
  • Zero edge at 87.5c against our own fair value of 88
  • $118K of volume makes any position expensive to establish or exit
  • This is a miss, not an opportunity, and treating it as one now would be chasing
4
ECB September Hike — effectively settled at 94.5c — YES
Sep 10, 2026·$92K on the leg·Confidence ★★☆☆☆ 4/10
↑ BUY YES+0.5pp
Market price
94.5%
Fair value
95%
Gap: +0.5pp
Verified at source today at 94.5c, up from 92c on Friday, with no change at 3.55c and a cut at 0.05c. There is nothing left to forecast here and we include it for one reason: it is the cleanest available evidence that the long-end repricing our main position depends on is a global phenomenon rather than an American fiscal story. The euro area is about to raise rates into an oil shock with Brent above $90 for most of a month, Germany's 10-year Bund is 3.2502% and close to its highest level since 2011, and Berlin has been issuing 30-year paper at yields last seen in the same year. When the ECB, the BoJ and the Bank of England are all tightening or holding at restrictive levels while sovereign issuance stays heavy, a US Treasury buyback of $4bn per operation across two maturity buckets for eight weeks is a liquidity gesture, not a change in the supply of duration. That is the whole argument for why we reversed Wednesday's fair-value cut on pos-017 and why today's adjustment was two points rather than fourteen. No trade: half a point of gap on a $92K leg is not an opportunity in any book, let alone one that already has four positions expressing a hawkish view.
▵ Bull case
  • Bund at 3.2502%, near a 2011 high, with 30-year issuance at matching levels
  • Brent above $90 for most of the month keeps the euro-area inflation impulse alive
  • Confirms that long-end repricing is global, which is the core of the pos-017 thesis
▿ Bear case
  • No edge at 94.5c and no position to take
  • $92K of volume on the leg
  • A quick Iran resolution would deflate the energy impulse in Europe faster than in the US
5
US 10Y Touches 5.00% Before 2027 — the internal consistency check — YES
Dec 31, 2026·$83K on the leg·Confidence ★★☆☆☆ 4/10
↑ BUY YES0pp
Market price
16.0%
Fair value
16%
Gap: 0pp
We include this leg because it is the check on our own headline position rather than a trade, and because if it disagreed with us we would have to say so. Our fair value on the 4.80% barrier is 72, derived from a 10.0bp distance, 93 sessions and about 4bp of daily volatility. Run the identical model out to 5.00%, which is 30.0bp away from today's 4.70%, and 2*(1-Phi(30.0/38.6)) gives roughly 43% raw. Apply a materially larger haircut than we use on the nearer barrier — because a 30bp move requires the repricing to continue rather than merely to finish, and because the buyback window plus any Iran resolution bite harder over a longer distance — and something in the mid-teens is where we land. The market is 16.0c on a 15 bid and a 17 ask. We are essentially exactly on the market, which is the outcome we want from a consistency check: it means our volatility assumption is not doing secret work to make the 4.8% leg look cheap. If we had computed 30c here we would have had to admit that our sigma was too generous and that our 72 on pos-017 was overstated. We did not, so we are leaving both alone. No trade, no edge, and no fifth ticket on the same underlying view.
▵ Bull case
  • Same model, same sigma, lands within a point of the traded market
  • Confirms the 4.8% fair value is not being manufactured by an inflated volatility input
  • Global long-end repricing would have to accelerate, not merely persist, to reach 5.00%
▿ Bear case
  • Zero gap means zero information beyond the consistency check itself
  • $83K of volume with a two-point spread
  • Correlation cap would block it even if a gap appeared