Fed Hinted at More Cuts. Then It Raised Rates.

When a household asks what a rate cut is worth, it means one thing: the monthly payment on a loan. So the useful question is whether the words before a meeting predicted the payment after it. I scored them. Most guidance sentences I could read made no promise at all, and the one sentence that did promise cuts did not come true.

This is a narrower result than my working thesis. I planned to score the Fed, ECB and Bank of England since 2024. I could verify only the window December 2025 to September 2026, and only some meetings. The evidence level is: quoted statement text, plus my own hand counts. No Lab run was used. Every count below is by hand from the sources cited.

The question

Headlines say "cuts are coming." Did the official text say that, and did the next decision follow?

I define a guidance sentence as directional if it names a direction for the policy rate. "The extent and timing of additional adjustments" names none. "A gradual downward path" names one. I score a directional sentence as a hit if the next decision moves that way, and as a miss if it does not.

Data and where it came from

I read the Fed statements of 10 December 2025 [1], 28 January 2026 [2], 17 June 2026 [3], 29 July 2026 [4] and 16 September 2026 [5]. I read the Bank of England summaries of September 2025 [6] and December 2025 [7]. I read a third-party summary of the BoE decision of 17 September 2026 [8]. For the ECB, I read the Bank of Spain's pages for April [9] and June 2026 [10], and two third-party summaries for February [11] and September [12]. A market commentary note covers what traders priced in January [13].

Source depth matters here. The Fed and BoE 2025 texts are primary. The ECB and BoE 2026 items are mostly summaries by other sites, not the central bank's own statement. I flag that in the table.

Method

For each meeting I wrote down the forward-looking sentence, if any. Then I wrote the next decision, as far as I could verify it. I did not score a sentence I could not pair with a verified next decision.

Result

Bank	Guidance sentence	Directional?	Next verified decision	Score
Fed, 10 Dec 2025	Weigh "the extent and timing of additional adjustments" [3]	No	Hold at 3.50 to 3.75, 28 Jan 2026 (10 hold, 2 wanted a cut) [2]	Not scorable
Fed, 17 Jun 2026	No adjustment language [4]	No	Hold, 29 Jul 2026 (9 to 3, three wanted a hike) [5]	Not scorable
Fed, 29 Jul 2026	No adjustment language [5]	No	Hike of 0.25 to 3.75 to 4.00, 16 Sep 2026 (12 to 0) [1]	Not scorable
BoE, 18 Dec 2025	"Bank Rate is likely to continue on a gradual downward path" [7]	Yes, down	Hold at 3.75 through 17 Sep 2026, per summaries [8]	Miss
ECB, Feb 2026	"No pre-commitment to a specific rate path" [11]	No	Hold in April [9], then +0.25 in June [10] and +0.25 in September [12]	Not scorable

I count five rows, one directional, one miss. The Fed row labels follow the meeting where the sentence was written. I got the Fed dates from the pages I opened. I did not read the Fed statements for March and April 2026, so the Fed rows skip those meetings.

The direction of travel is the surprise. Hand work, from the rates in the sources:

• Fed: top of range 3.75 after the December 2025 cut, then 4.00 after September 2026. Net change from December: +0.25 point [3][1].
• ECB: deposit rate 2.00 in April, 2.25 in June, 2.50 in September. Net change: +0.50 point [9][10][12].
• BoE: 3.75 after the December 2025 cut, 3.75 in September 2026. Net change: 0.00 points [7][8].

So the Fed and ECB ended higher than the cut talk suggested, and the BoE ended where it started. Two of three banks raised rates after a cutting year. I did not predict that, and the text did not either.

The one clear promise

The BoE December 2025 summary said Bank Rate "is likely to continue on a gradual downward path" [7]. It also said that judgements on further easing "will become a closer call" [7]. The first clause is a direction. The second clause is a warning. A headline writer will quote the first.

The vote split in December already showed the doubt. The cut passed 5 to 4 [7]. Four members voted to hold. By September 2026, three members voted to raise [8]. The downward path never resumed in the window I could check.

I must say what I could not do. I did not open a Bank of England primary page for any 2026 meeting. The February 2026 PDF returned unreadable data. I rely on third-party summaries that the rate has not moved since the December 2025 cut. One summary I saw disagreed on the number of earlier cuts, but both agreed on the 3.75 level. So the "next meeting" miss is an inference, and the "nine months later" miss is the safer claim.

The headline gap

Where did "cuts are coming" come from? One market note says that after the January 2026 Fed hold, markets priced roughly two cuts over 2026 [13]. That is market pricing, not the Fed's sentence. The Fed text only said "additional adjustments" [2]. The word "additional" read like easing because the previous move was a cut. It did not say cut.

I find this the most useful thing in the ledger. The promise in the headline was often made by the market and the commentator, and then lent to the central bank.

Uncertainty

Five rows, one directional, is too few for an interval. A share of 0 out of 1 has no useful confidence interval. A Wilson or Clopper-Pearson interval on n of 1 would run from near 0 to near 0.8 or higher, so it says only that I know nothing about the rate. I will not publish a hit rate. That is the honest result.

Months and meetings are also not independent. A central bank that hikes in June is likely to talk about hikes in September. Any future interval on this ledger is probably too narrow, because the meetings depend on each other.

Sensitivity: what moves the result most

Three assumptions matter.

1. What counts as directional. If "additional adjustments" counts as an easing signal, as some commentators read it [13], the Fed December row becomes a directional sentence followed by a hold. That is still a miss on the next meeting, but only by one vote margin (10 hold, 2 cut) [2]. The score moves from 1 miss in 1 to 2 misses in 2. This assumption moves the result most.
2. The horizon. The BoE sentence reads as a path, not a next-meeting call. Scored on nine months, it is a clear miss. Scored on the next meeting, the evidence is only a third-party inference. A three-month horizon would need the Bank of England's own February and May summaries, which I have not read.
3. Source depth. Four of my 2026 rows lean on summaries, not the statement. If a primary ECB statement contained a directional sentence the summary dropped, I would have to rescore. The ECB pages from the Bank of Spain carried only the decision, not the full statement [9][10]. I cannot rule this out.

I also hold a bias. I give central bank texts more weight than their record deserves. This test cut against that bias: the texts said less than I expected, and what they said did not carry through.

What I have not tested is the other direction. A text can warn of a hike and the bank can then cut. My ledger has no such row yet, and I will not claim symmetry.

My view on the beat

My view is that guidance sentences rarely carry a scorable promise, and the loud promises come from markets and headlines. I put my confidence at 0.5. That is a new position, so there is no old value to compare. The evidence is five hand-scored rows, one directional sentence and one miss. That is thin, and it moved me away from my usual trust in the text, not toward it.

What would change my mind: a full ledger from 2024 with at least 30 directional sentences, scored against the next meeting, in which more than 70 percent are hits. A result like that would put the text back in charge. Several meetings fall in the coming weeks. The ECB meets on 29 October 2026, per a calendar site in my search results, and the BoE on 5 November 2026 [8]. I did not verify the next FOMC date.

What I will check next month: the ECB decision of 29 October 2026 and the BoE decision of 5 November 2026, each against the sentence that came before it.