Fed Agent · The ledger resolves 16 September 2026
On the board · 2026-09 meeting
Hold75%
Hike 25bp25%
Recorded 1 August 2026 on data as known on 1 August 2026, frozen from that moment. It resolves on 16 September 2026, and not before: a prediction that can be edited after the fact is not a prediction.
The record
Modal callHold
Resolves16 September 2026
Expected dissents0.75
Modelanthropic.claude-opus-5
Dissent watch, named in advance

Bowman, Barr, Hammack

Status

Not scored. Nothing in this ledger has ever met a real meeting. Every accuracy claim about this machine today has no evidence behind it, and that stays true until the row below has numbers in it.

The two yardsticks, with their own dates

A prediction is only as meaningful as what it is measured against, and both baselines here are older than the prediction itself.
market 71% probability of a hike OIS-implied probability, JPM OIS monitor dated 2026-07-03 as of 3 July 2026 · four weeks before the prediction
published consensus modal call: Hold broker consensus, coverage through 2026-07-11 coverage ends 11 July 2026 · before the 29 July dissents
Why they are not being refreshed

Changing a baseline after the fact is moving the yardstick, and a prediction whose yardstick moves is not a prediction. The rule adopted instead is procedural: when the meeting resolves, both benchmarks are reported together with their as-of dates, and the baselines are refreshed before anything new is recorded. The direction of the error is stated rather than corrected - if the true market price on 1 August sat closer to a hike than the 3 July monitor did, the model was further from the market than this record says, not nearer.

How it will be scored

Two rules, because they answer two different questions. Both are run by ledger.py score on the frozen record, not on a re-reading of it.
Brier score · were the probabilities calibrated

The squared error summed across the whole action space, not just the call that turned out right: a confident right answer and a hedged right answer are not the same thing, and a distribution that hedged everything should not score like one that committed. Hold 75 / hike 25 scores 0.125 if it holds and 1.125 if it hikes. Lower is better. Below 0.25 beats always saying fifty-fifty, which is the only bar this machine has to clear to have said anything at all.

Dissent precision and recall · did it name the right people

The watch list is three names, written down in advance. Precision is how many of those three actually dissented; recall is how many of the actual dissenters were on the list. They are reported separately on purpose: naming everybody would give perfect recall and worthless precision, and naming nobody would do the reverse. This is the part of the output a rates market does not produce, so it is the part worth being measured on.

Beating the market is a separate line, filled in by hand

The engine does not grade its own comparison. The market benchmark is entered with its date, the outcome is entered with its date, and whether one beat the other is written down as a sentence rather than computed into a flattering number.

Scored history

Every meeting this engine has been graded against, once any have been.
MeetingPredictedActual BrierDissent P/RBeat market
No scored predictions yet. The first one lands on 16 September 2026.
What the first score will and will not settle

It retires exactly one sentence - “never scored” becomes “scored once”. One meeting is not a track record either way, and a right answer for a wrong reason will be published as such: the two checks below already name the mechanism that would make a correct hold suspicious.

Two checks that did not wait for the meeting

A rate call takes until September to grade. A persona can be graded the day the person speaks, and twice now it has been.
28 August · the Jackson Hole keynote

The Chair persona was distilled on 1 August from documents that existed then; the keynote did not exist until 28 August, so the model could not have read it. 5 substantive matches, of which the strongest is arithmetic rather than phrasing: the model counted the overshoot to “month 65 or 66” and he said “65 months”. Against that, one pillar failed outright. The model had the Chairman holding because the market had already delivered the tightening; he said in the keynote that he would be hard pressed to describe financial conditions as restrictive at all. That is not a shade of difference - it is the mechanism the hold rested on, gone.

31 August · the first reading overturned

Three days later the first reading was overturned - not the observation, the cause. The model did not invent the pillar. It is the Chairman's own mechanism, from his 29 July press conference, and it was in the corpus the card was built from. Two errors, both of them the kind that survives a spell-check: A hawkish argument was used as a dovish one - the sentence he actually said was vouching for transmission, not explaining why he could wait. And A change was promoted into a level - on 29 July he was describing the intermeeting change in conditions, while on 17 June and 28 August he was describing the level. The model read one as the other.

What that does to the number on the board

Nothing. The record for September is already closed, and editing it because new information arrived would mean it was never a prediction. What the re-diagnosis changes is the reading, and the reading tilts further toward a hike than the first one did. Both documents are reported alongside the result when the meeting resolves; if the outcome is a hike, this is the ready-made explanation of why the model was wrong.

Standing boundaries · what this ledger does not license you to say
01Never scored. Any statement about this engine's accuracy is unsupported until a meeting has been recorded, resolved and scored end to end.
02The sample of Warsh-chaired meetings is two. Persona uncertainty is real and the distribution has to stay wide.
03This does not try to beat the futures market's binary pricing. The product is disagreement detection, not the level.
04Member agents have a natural tendency to collapse toward consensus. Information isolation between memos is the defence, and it has not been validated against a retro-run of the June or July meetings.
data as of 8 September 2026 · deepseek-v4-pro · one full committee run ≈ ¥1.5, calibrated · Never scored against a real meeting — first scoring point 16 September 2026 Engine record: PREDICTION-LEDGER.json · entry 2026-09::2026-08-01