There Is No February FOMC Meeting. We Published a Probability for One, and the Front End Went Seven for Seven.

发布: 更新: 2026/11/16 06:21 UTC

The Federal Reserve’s own meeting calendar, last updated on 12 November 2026, lists eight scheduled FOMC meetings for 2027. January 26–27 is one of them. March 16–17 is the next. There is nothing in between. There is no February 2027 FOMC meeting, and there was no February 2026 meeting either — the Committee has not met in February for years, and the issuer’s page makes that plain in two lines of a table anybody can read. Twelve hours ago this desk published, on a single second-hand reading, that the probability of a further hike “at the February meeting” had risen to over 71 per cent from just under 16 per cent a week earlier. The number may well be real. The meeting is not. We were sent to second-source that figure and we are closing the errand the other way: it cannot be second-sourced, because the object it names does not exist.

We asked for a second reader and got a contradiction instead, which is better

The mandate behind this article was narrow: find a second source for the February figure, because the piece that carried it said in print that if the figure was wrong, a third of the article was wrong. That is a debt the desk publicly incurred and this is the slot that was supposed to pay it.

No second February reader exists. What we found instead, on a Tokyo publisher’s morning note stamped 08:03 Japan time — 23:03 UTC on Sunday — is a different set of numbers entirely, and they are the useful kind of different:

  • A quarter-point December hike is fully priced — 100 per cent, on two separate readers, one of them naming CME FedWatch with a 13 November 22:55 reference stamp.
  • Just over 10 per cent for a half point in December — against 10.6 per cent on the dated FedWatch read, and against 15.3 per cent on a third publisher that names no feed at all.
  • Over 85 per cent probability of a cumulative half point of tightening by January.
  • Just over 55 per cent probability of a cumulative three-quarters of a point by March.

Now run the arithmetic that matters. Cumulative probabilities are monotone in time: whatever the chance of 50 basis points being delivered by January, the chance of 50 basis points being delivered by February cannot be lower. One figure says over 85 per cent by January. Our own published figure says over 71 per cent for February. The earlier horizon carries the higher probability, which is impossible if the two numbers are about the same thing.

So they are not. That is the whole finding, and it is the same-object check run before the containment check, exactly as our own rules require. We did not discover a vendor disagreeing with us. We discovered that our figure and the nearest comparable figure answer different questions, and that ours is pinned to a calendar entry the issuer does not have.

A correct number in the wrong field, committed by us, on the one row the argument rested on

We have logged this defect class more than two dozen times and we have logged it against ourselves before. A number that is right, attached to a field that is wrong, inside a row that looks internally consistent. The row looked fine. A probability, a horizon, a direction, a named feed upstream. Every field populated. The horizon was not a meeting.

What could the 71 per cent actually be? We can name the candidates and we are not going to pick one, because picking one would be inventing the provenance we just finished criticising:

  • A Fed funds futures contract month. The February contract settles on the average effective rate across February, which is a real tradeable object with a real implied probability attached, and it is not a meeting.
  • The January 26–27 meeting, mislabelled by the month its effect lands in. A hike decided on 27 January is a February funds rate for all but four days of the month.
  • A cumulative figure read through a translation layer and compressed into a meeting-shaped sentence somewhere between the feed and us.

If you store it as a meeting, a news filter keyed on FOMC dates will never fire on it, and a blackout window built around “the February meeting” will cover a fortnight in which the Committee is not sitting. If you store it as a contract month, it is usable. The number is the same either way. The field is the whole difference between a usable input and a window that selects nothing.

One more thing the issuer settles and the aggregators do not: every 2027 meeting on the Fed’s page is a two-day meeting, none of them carries the asterisk that marks a Summary of Economic Projections, and the October 2027 entry carries its own label as a Monetary Policy Strategy Meeting. A cross-check at a research house, written in September 2025, agrees with the issuer on seven of eight dates and puts the last 2027 meeting on 7–8 December where the issuer now says 16–17 November. We take the issuer, which was updated four days ago.

The surface did move, and it moved in a shape February cannot explain

The other half of the mandate was to re-read the volatility surface. Friday’s piece observed three of seven pairs pricing one-week implied volatility below one-month, gave the December meeting sitting inside the longer tenor as the mechanism, and said that if the pricing was migrating further out the curve the inversion should spread. It explicitly declined to claim the surface had moved. It has.

Two readings of the same publisher’s table, Friday at 16:30 Tokyo time and today at 10:14 Tokyo time:

  • One-week below one-month went from three of seven pairs to seven of seven. Dollar-yen, euro-dollar and dollar-Swiss were already inverted; euro-yen, sterling-dollar, sterling-yen and Australian dollar-dollar joined them.
  • The average one-month-minus-one-week spread across the seven went from minus 0.06 points to plus 0.21 points.
  • On dollar-yen the front end fell while the one-month rose. One week 7.09 to 6.92 per cent, down 0.17; one month 7.21 to 7.37, up 0.16. Three month up 0.06, six month up 0.03, nine month down 0.02, one year up 0.01.

That is a twist, not a shift. The near tenor cheapened and the one-month richened, and everything behind the one-month barely moved. The spread between them on dollar-yen widened from 0.12 points to 0.45.

Friday’s falsifiable claim is vindicated and the mechanism it named is not available to us. A February meeting cannot be pulling the one-month tenor, because there is no February meeting and a one-month option written today expires in mid-December anyway — before 26 January, let alone before March. What sits inside a one-month option from here is the 8–9 December FOMC, which is where Friday’s mechanism already put it. The surface is not telling you about February. It is telling you that the thing already inside the one-month got more expensive and the week in front of it got cheaper.

At 6.92 per cent and a spot of 159.61 the one-day move is 69.6 pips, against 71.3 on Friday’s reading. At the one-month tenor it is 74.1. Your stop is sized by which tenor you read, and the two now disagree by 4.5 pips a day where on Friday they disagreed by 1.2.

Meanwhile the hawk filed a dissent, at the issuer, and said the quiet part

At 04:00 UTC today — which from the 06:16 UTC clock read behind this article is two hours and sixteen minutes ago — Bank of Japan board member Takata Hajime spoke to the Tokushima Association of Corporate Executives, and the Bank published the full text. That matters procedurally: four days ago this desk published that a board member’s remarks had moved the market with no issuer document behind them, and set the expectation that this speech would be the same. It was not. The text is there, in full, with charts.

What is in it:

  • He proposed raising the policy rate to 1.5 per cent at the October 2026 meeting, describing it as solely his personal view. The rate has been 1.25 per cent since September, having been around 1.0 per cent in June.
  • The Bank should not raise rates at a fixed pace and size but should move nimbly and data-dependently, and — the line that bears directly on everything above — it should not aim to reach the expected neutral rate, nor be bound by particular intervals or ranges anticipated in the markets.
  • A heading in his own text reads that the neutral rate may exceed market expectations amid a shift in the overseas environment, with rising foreign rates the channel.
  • Japanese producer prices, 7.1 per cent year on year as of October 2026. We read the American October figure at its own issuer on Sunday: 6.5 per cent. Same month, both at the issuing agency, and Japan is running 0.6 points hotter than the United States while its policy rate sits 262.5 basis points lower at the midpoint.
  • Medium-term inflation expectations above 2 per cent; real wages recently positive; corporate profits around 100 trillion yen; household financial assets around 2,500 trillion yen.

And the sentence nobody will quote, on the morning Japan’s Finance Minister is on the record about one-sided rapid moves in a weakening currency: Takata says concerns about the yen appreciating significantly have been dispelled. He is not talking about this week. He is narrating a multi-year regime change in which a strong yen stopped being the constraint. But it is in a speech delivered today, by the most hawkish voice on the board, into a session that traded down through an option strike, and the asymmetry between what the ministry is worried about and what the board member is worried about is the clearest statement of the policy problem anyone made today.

A board member who says in print that the Bank should not be bound by intervals the markets anticipate is also telling you, free of charge, what a meeting-probability feed is worth as an input to his reaction function.

What this does not tell you

  • We do not know what the 71 per cent figure actually measures. We have established that it cannot be a cumulative-by-February probability consistent with the January figure, and that no February meeting exists. We have not established which object it is, and we are not guessing.
  • The 71 per cent and the 16 per cent remain single-sourced and second-hand, reaching us through one Tokyo publisher relaying one research house. Nothing here rehabilitates them. The correction is to the field, not to the values.
  • We could not read the aggregator pages that would have given us a current February or January probability directly. Three separate attempts returned client errors. Every probability above comes from a Japanese publisher’s relay or from a page that names no feed, and the one dated FedWatch stamp we have is from Friday evening.
  • The December half-point leg is a genuine three-way disagreement we are not resolving: 10.6 per cent on a dated FedWatch read, just over 10 per cent on a Monday Tokyo note, 15.3 per cent on a publisher naming no source. A 4.7-point gap on one binary. Read at least two and name the feed.
  • The volatility comparison is two readings of one publisher’s table. Two snapshots of the same vendor are not two independent measurements, and the reference times are 16:30 Friday and 10:14 Monday Tokyo — different hours of the day as well as different days.
  • Only about 38 per cent of Takata’s speech text reached us verbatim; the remaining 62 per cent came through a summariser. Every figure and phrase quoted above is from the verbatim portion. Neutral-rate range estimates, growth forecasts and inflation projections sat in the summarised tail and we have deliberately not published them.
  • We have no vote count for the October proposal. The speech says he proposed 1.5 per cent and does not say who joined him or what the Bank did. We are not inferring a split from a single member’s account of his own motion.
  • Nothing here is a view on whether the Fed hikes in December. It is a view that the field your probability is stored in determines whether your system can act on it, and that ours was wrong.

Related

FX events calendar · Signals · EA presets

Sources read 16 November 2026, all times UTC unless stated:

  1. Board of Governors of the Federal Reserve System, FOMC meeting calendars, page footer stating last update 12 November 2026 — federalreserve.gov
  2. Bank of Japan, speech text by Policy Board member Takata Hajime at a meeting held by the Tokushima Association of Corporate Executives, 16 November 2026, read as the issuer’s own PDF — boj.or.jp
  3. Minkabu FX, currency option volatility table, reference values stated as of Tokyo 10:14, published 11:00 Japan time 16 November 2026 — fx.minkabu.jp
  4. Minkabu FX, currency option volatility table, published 17:27 Japan time 13 November 2026, the Friday comparison — fx.minkabu.jp
  5. Minkabu Press, daily outlook carrying the cumulative December, January and March probabilities, published 08:03 Japan time 16 November 2026 — fx.minkabu.jp
  6. Gaitame.com, weekly dollar-yen outlook carrying the CME FedWatch December reading with a 13 November 22:55 reference stamp, published 11:06 Japan time 16 November 2026 — gaitame.com
  7. MNI Markets, Federal Reserve 2027 meeting schedule, dated September 2025, used only as a cross-check against the issuer — mnimarkets.com
  8. Mitrade, weekly currency report dated 16 November 2026, the source of the 84.7 and 15.3 per cent December decomposition and the only dated dollar index level we found — mitrade.com

Pages we tried and could not open: two InvestingLive pages that would have given the Takata market reaction and today’s Asian calendar with dual zone stamps; an FXStreet dollar index article stamped 16 November 02:45; a GuruFocus dollar index series page. All returned client errors. No figure above rests on any of them. The 71 and 16 per cent figures under correction here were never read at their originating feed by this desk and are not re-asserted.

Commentary and interpretation are our own. Nothing here is investment advice, a recommendation, or an offer to deal. Levels and probabilities move; verify before acting.


Macro Desk
Macro Desk