Figure 1 · Two venues, one question: the expected change in the policy rate at the July 28-29 FOMC meeting, in basis points. Polymarket’s five-rung decision ladder collapsed to its expected change, beside the same quantity implied by CME 30-day Fed Funds futures, across the 50 paired sessions from May 8 to July 22, 2026, aligned at the futures close. The pale band is the conversion’s own resolution. This describes how two published prices moved before the meeting; neither is scored.
Two Prices for the Fed
A retail prediction market and the institutional fed funds futures strip put a price on the same event: what the Fed would do at its July 28-29 meeting. Over the run-up to it, May 8 to July 22, 2026, the two moved together day after day, while Polymarket sat a small, steady distance below. Neither venue is scored for accuracy: accuracy is not what this measures.
The finding
Two very different places price the same event. Polymarket runs a five-rung ladder on each FOMC meeting; a few miles of fibre away, CME’s 30-day Fed Funds futures have priced the same meetings for decades. For the July 28-29 decision we set the two side by side as one number each: the expected change in the policy rate at that meeting, in basis points.
They tracked. The rank correlation of the two series’ day-over-day changes was +0.76 (95% interval +0.57 to +0.91). Counting instead, on the 26 sessions where either venue moved beyond the conversion’s own resolution, both moved the same way on 21. A retail crowd and an institutional futures market re-priced the same meeting on the same days, in the same direction. What links them is not tested here.
Two very different venues, one meeting, moving the same way.
Two Prices for the Fed · Vera Research
What the two prices are
You cannot read a probability off a futures price by looking at it; getting there takes an explicit chain of settlement and day-count assumptions. So the comparison leans on as few as it can, reducing each venue to its own expected change in basis points, with no distributional assumption on the futures side at all.
One thing belongs at the top. This note does not score either venue for accuracy. Accuracy is not what it measures, and scoring it is outside what we publish. What it measures is the distance between two published prices over a window that closed before the meeting, and that is what it reports.
The offset, described
Alongside the agreement, a second fact. Across the 50 paired July sessions Polymarket’s expected change sat about 2.4 basis points below the futures’, and it stayed there: the difference held beyond twice the conversion’s own resolution for a run of 32 sessions in a row, with only 7 of the 50 closing inside that band. It is more than three times that resolution, so it is not an artefact of the conversion’s grid.
This note reports that difference and does not explain it. At least three ordinary things could set a small, steady distance between two correctly priced venues, and this window cannot separate them: the two instruments settle on different objects, Polymarket on a change in the target range and the futures on the realised funds rate that floats inside it; the futures carry a premium for the hedgers who hold them; or the two simply saw the meeting a little differently. The distance is measured, not explained, and the note does not say which way it would close or treat it as something to act on.
The news moving money, markets, and the world—before your day starts.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Further out, and the limits
The near meeting is the clean one, and the only one on a full window: September rests on 31 paired sessions from June 8, October on 21 from June 23, both ending July 22. September co-moved as well, though less tightly, at a rank correlation of +0.48 with the two moving the same way on 11 of 14 sessions, and carried a wider difference, about 5 bp, in the same direction. October is a null: its day-over-day correlation was +0.10 and did not clear significance, the futures-implied number having barely moved beyond its own grid, and the two moved the same way on 4 of 9 sessions, fewer than half and the weakest of the three. Both are reported at the same prominence as the rest.
Three meetings, inside one spring-summer window in which the target range never moved, is not a sample, and there is no easing cycle anywhere in this data. The one consistent thread, Polymarket below the futures at all three, is suggestive and no more. What the note stands behind is narrow and sturdy: two separately quoted prices for the same event tracked each other closely, and carried a small persistent difference that it describes and leaves open.
The takeaway
- This note does not score either venue for accuracy. Accuracy is not what it measures, and scoring it is outside what we publish. The three meetings (July, September and October 2026) are priced here over windows that closed before each one.
- They track. On the July meeting the two series’ day-over-day changes had a rank correlation of +0.76 (95% interval +0.57 to +0.91), and on the 26 sessions where either venue moved beyond the conversion’s resolution, both moved the same way on 21.
- A small, steady difference. Polymarket’s expected change sat about 2.4 bp below the futures’ across the 50 July sessions, holding beyond twice the conversion’s resolution for 32 sessions in a row. It is described, not explained: a settlement difference, a hedging premium or a genuine difference in view could each produce it.
- Weaker further out, on shorter windows. September, 31 sessions from June 8, co-moved less tightly (+0.48, same direction 11 of 14). October, 21 sessions from June 23, showed no measurable daily co-movement (+0.10, not significant, same direction 4 of 9). Reported at equal prominence.
- What it is: a description of two published prices for one event over a past window. Not a signal and nothing to act on. The dataset is Vera.
The difference between the venues is measured, not explained, and the note does not say which way it would close. Neither venue is scored for accuracy, which is not what this measures. Every figure reproduces from the committed analysis bundle.
Method and data
Built from the committed two-prices-fed analysis bundle: Polymarket ladder prices (gloria pm_prices, read-only), CME ZQ contract settlements (Yahoo Finance) and the effective funds rate (FRED), aligned at the futures close. Every window ends July 22, 2026 and each meeting has its own: July 50 paired sessions from May 8, September 31 from June 8, October 21 from June 23. Each venue is reduced to its expected change in the policy rate at the meeting, in basis points. Every figure reproduces from the bundle, and a re-run of the analysis matches its committed output byte for byte. A descriptive measurement over windows that closed before the meetings they price: neither venue is scored for accuracy.
Odds shown are each market’s own price at the time Vera flagged the story, stated as fact, not advice. Nothing here is a recommendation to buy, sell, or trade any market or asset. Vera and Crypto Briefing are not registered investment advisers.