Polymarket market

Will the Fed Hike–Pause–Hike in the next three decisions (Sep–Oct–Dec)?

Uncategorised · Open · scheduled end Dec 09, 2026 · 23:59 UTC

← All tracked markets

Two separate probabilities

Polymarket-implied "Yes"76.0%

Market-derived: Polymarket's own last-traded price for the first outcome.

NVIDIA Kumo Relational estimate of "Yes"26.3%

Independent method: a supervised model over pre-resolution market microstructure.

NVIDIA Kumo Relational

Model estimate

A supervised relational model asked how this market eventually resolves. Its context is settled markets described only by what was observable a fixed horizon before their resolution.

ready An estimate from the live endpoint is cached below.

Model estimate of "Yes"26.3%

The model's probability for the first outcome resolving true.

Decision: Yes resolves false. Probabilities: true 0.2628 · false 0.7372.

Provenance

Model: kumo-relational
Endpoint: /v1/structured-data/nvidia/kumo-relational/predictions
Answer generated: Sep 23, 2026 · 16:30 UTC
Prediction anchor: Sep 23, 2026 · 16:30 UTC
Horizon before resolution: 24 h
Context rows (settled markets): 120
Prediction rows: 847
Pre-anchor price points: 13666
Trade-flow rows: 2
Order-book rows: 0
Request id: pred_14fe7a00c568

Features supplied

px_first
px_last
px_mean
px_min
px_max
px_std
px_delta
px_points
px_span_secs
hours_to_end
poly_trade_flow(window=3600s)
poly_trade_flow(window=21600s)
poly_trade_flow(window=86400s)
poly_book_snapshot(band=0.02,side=bid,index=0)
poly_book_snapshot(band=0.02,side=ask,index=0)
poly_book_snapshot(band=0.05,side=bid,index=1)
poly_book_snapshot(band=0.05,side=ask,index=1)

Limitations

Context is 120 resolved markets scored 24 hours before resolution, so the model sees no outcome, final price or final volume from the market it scores.
No context row currently carries book features, so the microstructure columns cannot yet influence the estimate.
The market price is Polymarket's own last-traded probability: market-derived evidence, not independent evidence.
Order books are only published as a live snapshot, so no historical book exists to learn from; book features are therefore absent from most context rows by construction, never zero-filled.
Trade flow counts only trades reported on the market's first-outcome token. Second-outcome trades are excluded rather than converted.
No accuracy, calibration or backtest figure is published: none has been measured on a held-out sample.
A prediction is the model's estimate of how the market eventually resolves, not a forecast of the next price move.

No explanation, accuracy or calibration figure is shown, because none is measured or supplied by the endpoint. The model's probabilities are its own output; they are not this site's opinion and not investment advice.

Price path

1087 observations · probability · source Polymarket public CLOB history (requested by token id) · Sep 16, 2026 · 15:00 UTC to Nov 01, 2026 · 06:37 UTC. The dashed line marks the prediction anchor at Nov 01, 2026 · 15:22 UTC.

Market microstructure

Public order book and public trade history, as published. These are the inputs the model is allowed to see.

Order-book snapshot captured Nov 01, 2026 · 06:37 UTC
Measure	Bid	Ask
Best price	0.7100	0.7200
Depth within 0.02	7343.79	109881.74
Depth within 0.05	8466.43	115066.44
Imbalance within 0.02	-0.8747
Imbalance within 0.05	-0.8629

Spread 0.0100 · depth bands are absolute price bands around the best quote on each side; a fetched book with an empty side is a real zero, while an absent snapshot is reported as missing. Snapshot age 31505 s.

Public trade history for the first-outcome token, anchored Nov 01, 2026 · 15:22 UTC
Window	Trades	Buy size	Sell size	VWAP	Net
Last 1 h	No trade was reported in this window — missing, not zero.
Last 6 h	No trade was reported in this window — missing, not zero.
Last 1 d	1	-0.00	22.50	0.7100	-22.50

Only trades the public Data API reports against this market's first-outcome token are counted, using the side it reports; second-outcome trades are excluded rather than converted. Window ends at Nov 01, 2026 · 15:22 UTC.

Every microstructure measure here is an observation, never a default: a book side the API did not publish, a window in which no trade was reported, or a snapshot that was never recorded is reported as missing, not zero. A zero on this page always means the API genuinely reported zero.

Comments and bookmarks

Comments are plain text from NPDev accounts. Bookmarks stay private to the account that saved them.

No comments yet. Be the first to say something.