PricingtheFed:Kalshi,Polymarket,andCMEFedWatch

In November 2025, Polymarket and Kalshi priced "no change at the April 29 Fed meeting" 20¢ richer than CME FedWatch did. The gap held for six weeks. By April 28 — the day before the decision — all three sources had converged to within 1¢.

That gap isn't the crowd beating the bond market. It's the two markets answering different questions. And once you see why, the divergence becomes more interesting, not less.

We compared how three independent sources priced the same Fed-meeting outcome across three consecutive 2026 FOMC decisions (Jan 28, Mar 18, Apr 29 — all "no change"):

• Kalshi — CFTC-regulated US prediction-market exchange, per-meeting binary decision markets
• Polymarket — onchain prediction-market venue on Polygon, per-meeting binary decision markets
• CME FedWatch — the standard rates-market implied probability tool, computed from 30-Day Fed Funds futures

Daily series, from each prediction-market's listing date through resolution. Comparable notional across the period: ~$12M Kalshi, ~$232M Polymarket, daily Fed Funds futures volume in the hundreds of billions.

Explore Prediction Markets data

1. Prediction markets ran richer than FedWatch for months ahead of each meeting. Both Kalshi and Polymarket priced "no change" consistently higher than FedWatch from roughly 3 months out, with the gap scaling with horizon. Peak gaps reached 5–15¢ on the January meeting, 8–14¢ on the March meeting, and 20+¢ on the April meeting in November–December. Same directional bias on all three.
2. Convergence in the final 2–3 weeks. As each meeting approached, all three sources converged within 1–2¢. The "no change" probability rose to 95%+ on every venue before resolution.
3. Kalshi and Polymarket tracked each other closely. Cross-venue spread was small (~1¢ on average) and switched sign over time. The basis between the two prediction markets is a separate story from the basis to FedWatch.

The three sources are not asking the same question.

Polymarket and Kalshi binary markets ask: will the Fed change rates at this specific meeting? — conditional on whatever rate is in effect entering the meeting.

CME FedWatch reports: what is the probability the terminal rate after this meeting is X%? — marginalized over the path of all intermediate meetings between today and the target meeting.

These coincide on the meeting date and when no intermediate meetings exist in between. They diverge in earlier windows when intermediate FOMC outcomes are still uncertain. The 20¢ April gap in November partly reflects this: Polymarket was pricing the April action only, while FedWatch was pricing the full path through December, January, March, and April. As the intermediate meetings resolved, the path uncertainty collapsed and the questions converged into the same question.

So the gap is real and observable, but it shouldn't be read as "prediction markets called the hold before FedWatch did." The two markets diverged for months because they were measuring different things, and converged at resolution because they finally weren't. Whether the residual gap — the part not explained by path marginalization — carries incremental information is the question this analysis opens, not closes.

The analysis required hourly OHLCV from Kalshi and Polymarket on a unified schema, joined with FedWatch data on the same time grid. The new dataset makes this comparison straightforward — without it, doing this requires reconciling three different vendor APIs and schemas.

The same methodology extends to:

• Cross-venue basis trading — sustained 1–2¢ spreads between Kalshi and Polymarket are observable, and the new data makes them tradeable as a continuous signal rather than a point-in-time snapshot
• Path-marginalization studies — comparing prediction markets to futures-implied probabilities, properly conditioning on intermediate event outcomes
• Event-driven research more broadly — any binary outcome priced on both venues plus a tradfi instrument can be studied with the same workflow

• Daily three-way series, Jan 28 meeting
• Daily three-way series, Mar 18 meeting
• Daily three-way series, Apr 29 meeting
• Monthly summary across all three