The December Repricing Happened Yesterday. One of the Three Baselines Says Nothing Moved.

Three baselines, and the one that names Thursday says almost nothing moved

Take the readings in the order their publishers stamped them, because the order is the finding.

• A Reuters-sourced metals wrap carrying a timestamp of 00:47 UTC today says traders see an 80 per cent chance of a December hike, attributed to the CME FedWatch tool. The word in the copy is “currently”, but the market content of that piece is Thursday’s US morning — its gold quote is explicitly 14:14 GMT Thursday — so the figure’s own as-of time is not printed anywhere on the page.
• A second Reuters-sourced FX wrap, stamped two hours later at 02:44 UTC, has the probability at “around 70 per cent” from “more than 80 per cent on Wednesday”, attributed to fed funds futures and to traders.
• A European desk note timestamped around 04:00 CET has it “near 70 per cent”, down from “about 83 per cent on Wednesday”, attributed only to “futures pricing” with no feed named.
• A fourth reader, undated beyond today, has it at around 70 per cent from “more than 70.9 per cent on Thursday”, again attributed to CME FedWatch.

Put those four in a line and they are not in conflict. They are a time series that two of the four have mislabelled. Eighty per cent survived into Thursday’s US morning. By Thursday’s close it was 70.9. This morning it is about 70. The entire repricing is Thursday afternoon’s, and the two pieces that reach for Wednesday as the comparison are skipping the day on which the thing actually happened.

That matters operationally and not rhetorically. A system that keys on “December odds collapse” as a Friday headline is reacting to a Thursday move it has already paid for. The same named feed — CME FedWatch, not two different vendors — is quoted at 80 and at 70 by the same wire service two hours apart. This is not a vendor disagreement of the kind this desk usually writes about. It is one series, read at unstated times, and the spread between the readings is entirely a function of when somebody looked.

And note what the level still is. A hike is priced at roughly seven chances in ten on the day after the broad US equity index fell 4.4 per cent to 7,353.73 and the Nasdaq 100 fell 6.4 per cent to 26,629.81. The rout took out about a tenth of the probability of a hike. It did not take out the hike.

The curve steepened on a 4.4 per cent equity day, which is the wrong shape

Here is the part worth more than the probability. The two-year Treasury yield fell about 10 basis points to 4.47 per cent on one reader and over 11 basis points to near 4.46 per cent on another. The ten-year fell about five basis points, to near 4.83 per cent on the second reader and 4.84 per cent on the first, which the first calls the lowest since 10 September.

Front end down ten or eleven, long end down five. That is a bull steepening of roughly six basis points, and an independent series confirms it: the published 10-year-minus-2-year spread sits at +0.38 percentage points after what the publisher describes as a one-day steepening of 6 basis points. Because that is a daily series, its “one day” is Thursday — the same session as the probability move.

A genuine growth scare does not do this. A flight to quality bids duration hardest and flattens the curve, because the thing being repriced is the terminal level of the economy. What happened instead was that the market took money out of the next two meetings and left the term premium alone. That is a near-dated inflation and policy repricing wearing an equity rout’s clothing, and the distinction decides whether you are short the front end or long the back.

Two other readings point the same way. Equity volatility went vertical — the VIX rose almost 70 per cent to 28.17 — while the MOVE index, the rate-volatility equivalent, stayed below 100. The bond market did not panic. And credit did widen: a high-yield spread index gained 16 basis points to 340, matching its cycle high since early 2025. So the stress is in equity and in credit, and the Treasury market’s response was to reprice the Fed rather than to buy insurance.

The driver is one year of inflation expectations, and it is a 25 basis point move

If you want the causal link rather than the correlation, it is in the inflation swaps, and the numbers are larger than anything in the yield complex. On a bank research reading, the one-year US inflation swap fell 24.9 basis points to a two-month low of 2.16 per cent, and the one-year euro swap fell 25.9 basis points to a two-month low of 3.10 per cent. Both in a session.

The reason is not mysterious. Brent fell nearly 9 per cent in the previous session and crude is down about 12 per cent on the week. A metals trader at High Ridge Futures put the mechanism plainly: weaker energy prices mean inflation fears subside and a lesser need for rate hikes down the road. Gold agreed, rising 1.57 per cent to $4,123.39 an ounce with a session high of $4,146.63, the highest since 20 October, after a 3.4 per cent jump the day before.

A 25 basis point fall in one-year inflation expectations should move the first two meetings and barely touch the ten-year, because an energy shock washes out of the year-ahead number and does not change the long-run path. That is precisely the curve shape printed above. The equity rout and the rate move share a date; they do not obviously share a cause, and a desk that attributes the front-end rally to semiconductors has the wrong explanatory variable in its model.

Against that, the hawkish case is still on the record. Cleveland’s Beth Hammack said policy needs to move to a restrictive stance. Euro-area inflation printed 4.0 per cent in October, a three-year high, with core at 2.5 per cent, and two Governing Council members have said in the last week that a hike next month is likely. The Bank of England raised to 4 per cent yesterday on an 8–1 vote with its own forecast reaching 4.9 per cent. Nobody in this complex is cutting.

We read the statement today, and it moves five things in our own file

On Wednesday this desk published an article about the Treasury refunding whose footer said, in terms, that we had not read the statement — Treasury’s own link for it returned a client error on every attempt, and every word of the new guidance sentence reached you through a wire. That page opened today. Marking our own work:

• The sentence is confirmed verbatim at the issuer. Treasury “anticipates maintaining nominal coupon and FRN auction sizes for the next several quarters.” The deletion of “at least” is real, the wire that reported it was right, and the rendering gap we flagged — one reader wrote “FRN”, another “floating-rate note” — resolves to “FRN”.
• Next week is confirmed and now has a settlement date. $58 billion of three-year notes at 1:00 p.m. ET Monday 9 November, $42 billion of ten-years Tuesday the 10th, $25 billion of 30-years Thursday the 12th, all on a yield basis, all settling Monday 16 November. In UTC each auction is 18:00. $125 billion offered to refund approximately $97 billion maturing on 15 November, which is roughly $28 billion of new cash — a figure we published from a secondary reader and can now stand behind.
• We were wrong about a date, in a caption. Our Related block said “the February refunding already dated 1 February 2027”. The issuer says the next quarterly refunding announcement is Wednesday 3 February 2027. Our footer was precise — it said the next financing estimates are dated 1 February — and the caption collapsed two events into one. February is a two-step, exactly as November was: estimates Monday, announcement Wednesday. The number was right for one object and wrong for the one it was attached to.
• The 7-year change is four months away, not now. We wrote that eight of twelve dates a year stop being new issues, which is right, and implied it was live. The issuer says the conversion begins with the March 2027 new issue, reopened in April and May, with new issues in March, June, September and December. The 7-year is listed at 44 for November, December and January. Any window keyed to a 7-year auction is correctly sized until March.
• There is a line we did not have. Treasury anticipates potentially issuing a short-dated cash management bill around the end of November, with reductions to shorter-dated bill sizes in December and increases across the bill curve in January. Buybacks of up to $70 billion in off-the-run securities across buckets plus up to $25 billion in the one-month to two-year bucket are confirmed at the issuer.

The trading point: $67 billion of ten-year and 30-year supply lands next week on the segment of the curve that did not rally yesterday. The front end, which did rally, is being sold in the smallest of the three sizes. If the inflation-swap move is right, the three-year on Monday is the auction with the best-priced tailwind and the 30-year on Thursday is the one carrying a term premium nobody repriced.

What this does not tell you

It does not tell you the probability is 70 per cent. Our best reconstruction of the sequence — 80 into Thursday morning, 70.9 at Thursday’s close, about 70 now — rests on four readings of which two carry no as-of time at all and one carries no publication time. We are confident about the shape and not about any single figure. Never quote any of these as “the market-implied probability”.

The ten-year has a seven basis point spread across readers and we did not resolve it. Two readers put it at 4.83 and 4.84 per cent. A third, a FRED series card for DGS10 marked only “7:00pm ET” with no date, shows 4.90 per cent and the same −5 basis point change. Same delta, different level, which usually means the two are reading different sessions. We used the 4.83–4.84 band and we are telling you the third reading exists rather than quietly dropping it.

It does not isolate anything. Thursday contained a semiconductor rout, a 9 per cent crude collapse, a Bank of England hike, jobless claims at 189,000, and productivity at a 3.6 per cent annualised rate with unit labour costs down 0.7 per cent. Attributing the front-end rally to the inflation swaps is our reading of a plausible mechanism, not a decomposition. We have not seen an event study and we are not pretending to one.

The signer is confirmed only from a page title. The refunding statement is published under the name of the Assistant Secretary for Financial Markets, Hunter McMaster. The statement body does not repeat the name. On Wednesday we refused to assert it off one machine-readable page; we now have two, which is better and is not the same as a signature.

And it does not tell you what payrolls does to any of this. The print is at 13:30 UTC, two hours after this note. One bank’s view, on the record this morning, is that recent data suggests labour-market improvement and that it continues in this report. A metals trader’s view, also on the record, is that a strong print would confirm an already highly expected December hike. Both of those are compatible with everything above.

Related

• FX event calendar — the 9, 10 and 12 November auctions at 18:00 UTC, settling 16 November, and the February two-step: estimates 1 February 2027, announcement Wednesday 3 February 2027.
• Signals — why a bull steepening and a flight to quality are different trades wearing the same headline.
• EA presets — news windows keyed to a 7-year auction that stays a monthly new issue until March 2027.