How well do Fed markets predict rate decisions? Kalshi’s favourite the day before has been right at all 14 meetings since 2025
We looked at Kalshi’s prices before every Federal Reserve decision from January 2025 to September 2026. The day before, the favourite was right 14 times out of 14, at an average price of 95%. A month out the market missed once, badly: September’s hike was priced at 29%. Volume has grown fifty-fold. The data, the misses and what it means for today’s meeting.
In this article
Kalshi's Fed markets have called every rate decision since the start of 2025, as long as you looked the day before. Across the 14 Federal Reserve decisions from January 2025 to September 2026, the outcome Kalshi priced highest at the close the day before the meeting was what the Fed did, every time. Its average price was 95%.
A month out, the picture is different. The market was wrong once, on September's hike, which was priced at just 29% thirty days before. With the Fed announcing today's decision at 2 p.m. ET, here is the record.
Key findings
Source: Kalshi market data: KXFEDDECISION events, daily closes; our analysis
- The day before, the market was always right. The favourite won all 14 times, at an average price of 95%.
- A week before, it was also always right, but less confident: the eventual outcome averaged 91%.
- A month before, it missed once. In mid-August, a September hike was priced at 29% and a hold at 72%. The Fed hiked.
- The most uncertain meeting it got right was July 2026, when a hold was priced at only 75% the day before and the Fed held.
- Volume has grown about fifty-fold, from 2 million contracts on the January 2025 meeting to 103 million on September 2026's.
The full data is in our CSV file.
Every decision
| Meeting | Decision | Price 30 days before | 7 days before | Day before |
|---|---|---|---|---|
| Jan 2025 | Hold | 89% | 99% | 98% |
| Mar 2025 | Hold | 98% | 98% | 99% |
| May 2025 | Hold | 74% | 92% | 97% |
| Jun 2025 | Hold | 87% | 97% | 98% |
| Jul 2025 | Hold | 81% | 95% | 97% |
| Sep 2025 | Cut 25bp | 72% | 82% | 91% |
| Oct 2025 | Cut 25bp | 80% | 96% | 98% |
| Dec 2025 | Cut 25bp | 68% | 93% | 97% |
| Jan 2026 | Hold | 87% | 97% | 99% |
| Mar 2026 | Hold | 92% | 98% | 99% |
| Apr 2026 | Hold | 97% | 99% | 99% |
| Jun 2026 | Hold | 97% | 98% | 99% |
| Jul 2026 | Hold | 82% | 77% | 75% |
| Sep 2026 | Hike 25bp | 29% | 54% | 87% |
Price of the outcome that happened, at Kalshi's daily close 30 days, 7 days and 1 day before the decision. Source: Kalshi public market data.
How the prices sharpen
Source: Kalshi market data; our analysis
Show the dataHide the data
| Value | |
|---|---|
| 30 days before | 80.9% |
| 7 days before | 91.1% |
| Day before | 95.2% |
The pattern is what you would expect from a market that absorbs information as it arrives. A month out, inflation reports, jobs data and Fed speeches are still to come, and the price of the eventual outcome averages about 81%. By the week of the meeting, officials are in their "blackout" period and most of the data is in, and the average reaches 91%. The day before, it is 95%.
The Brier score, which measures how far forecasts are from outcomes (0 is perfect), averages 0.014 the day before a meeting across the 14 decisions. One meeting, July 2026, accounts for most of that.
Much of the time, this is because the Fed tells markets what it will do. Policymakers prefer not to surprise anyone, and they shape expectations through speeches and the statement. A 99% price the day before is often the market reading the Fed's own signals correctly.
Holds versus moves
The market has found it easier to call a hold than a move:
| Type of decision | Meetings | Average price of the outcome, 30 days before | Day before |
|---|---|---|---|
| Hold | 10 | 88% | 96% |
| Cut or hike | 4 | 62% | 93% |
Source: Kalshi public market data; our analysis.
A month out, the four meetings where the Fed moved were priced at only 62% on average, against 88% for holds. By the day before, the gap had almost closed. That fits how the Fed works: it rarely moves without signalling first, but the signals tend to come late, in the final speeches and data releases before the blackout.
The three cuts of 2025 show the pattern. September's first cut was at 72% a month out and 91% the day before; October's at 80% and 98%; December's at 68% and 97%. Each time, the price of the cut climbed as the meeting approached.
The miss: September 2026
Source: Kalshi market data: KXFEDDECISION-26SEP, daily candles
Show the dataHide the data
| Hike 25bp | Fed maintains rate | |
|---|---|---|
| 18 Jul | 37% | 62% |
| 28 Jul | 55% | 42% |
| 7 Aug | 35% | 64% |
| 17 Aug | 29% | 72% |
| 27 Aug | 30% | 70% |
| 1 Sep | 61% | 39% |
| 6 Sep | 50% | 49% |
| 11 Sep | 78% | 22% |
| 15 Sep | 87% | 12% |
September's market was the most volatile of the period. A hike was briefly the favourite in late July, fell back to about 30–35% through August, then jumped above 60% at the start of September before settling at 87% the day before the meeting. The Fed raised rates by a quarter point, its first hike since 2023.
Thirty days before, the market had a hold at 72%. That is the only time in the 14 meetings that the 30-day favourite was wrong. It is also a reminder of what a 72% price means: the other outcome happens more than a quarter of the time.
The near miss: July 2026
July is the meeting where the day-before favourite was least secure. A hold was at about 94% in mid-July, but the price of a hike rose in the final days, and the day before the meeting a hold was priced at only 75%, with a hike at about 26%. The Fed held.
In hindsight, July was the market sensing the turn that came in September. In Brier terms, it was the worst day-before forecast of the 14 because the market gave real weight to an outcome that did not happen; in practical terms, the favourite still won.
What could make the record misleading
Fourteen correct calls in a row sound impressive, but most were easy: eleven of the fourteen outcomes were priced at 97% or more the day before. A forecaster who simply repeated the Fed's own guidance would have done nearly as well. The real test of a market is the hard cases, and there were only a few: July 2026, when the market gave a hike a one-in-four chance that did not happen, and September 2026, when it reached the right answer only in the final weeks.
There is also a limit to the period. Kalshi's Fed markets have only been large since 2025, a stretch without an emergency move or a financial crisis. How they would behave in a sudden shock, like the emergency cuts of March 2020, is untested.
Why the volume matters
Source: Kalshi market data: KXFEDDECISION events, total volume
Show the dataHide the data
| Value | |
|---|---|
| Jan 2025 | 2.0m |
| May 2025 | 41.3m |
| Sep 2025 | 91.8m |
| Jan 2026 | 36.0m |
| Jul 2026 | 68.8m |
| Sep 2026 | 103.2m |
The market that called these decisions in early 2025 was tiny: the January 2025 decision traded about 2 million contracts. By September 2026 it was about 103 million. The busiest meetings are the uncertain ones: September 2025, the first cut of that cycle, and September 2026, the first hike, both drew far more trading than the quiet holds in between.
A bigger market is harder to push around and attracts more professional traders, including those who also trade fed funds futures. That is one reason the prices track the CME's futures-based estimates so closely. The October market had traded about 19 million contracts by Tuesday's close.
What it means for today
On Tuesday's close, Kalshi priced a hold today at 96% and a quarter-point hike at 5%. On this record, a 96% favourite the day before has never lost. That does not make today certain, but it puts the burden of proof on anyone betting against a hold.
The more interesting market is December, where a quarter-point hike was priced at 74% on Tuesday's close and a hold at 24%. Thirty days before a meeting, the record above says the favourite is right most of the time but not always, and the price usually moves a lot as the data arrives. How to trade both markets, and what each contract pays: how to bet on Fed rate decisions. The schedule and background: next Fed meeting. Live prices: October and December on Kalshi.
Method
- Data: Kalshi's Fed decision events (KXFEDDECISION) for the 14 FOMC meetings from January 2025 to September 2026, daily candles for the 60 days before each decision. Older events were read from Kalshi's historical market archive.
- Prices: the daily close 30 days, 7 days and 1 day before each decision. The day-before price is the last close before the decision day.
- Accuracy: the favourite is the outcome with the highest price at that point. The Brier score sums the squared errors across all five outcomes of each meeting and averages over meetings.
- Limits: 14 meetings is a small sample, and most of them were easy calls. The record shows the market is reliable when the Fed has signalled its intentions; it says less about genuine surprises, which are rare.
Key points
- Day before: the favourite won all 14 Fed decisions since January 2025, at an average price of 95%.
- A month out: 13 of 14, with September 2026's hike priced at only 29%.
- Sharpening: the outcome's average price rose from 81% to 91% to 95% as the meeting approached.
- Volume: from 2 million contracts in January 2025 to 103 million in September 2026.
- Today: a hold at 96% on Tuesday's close; December's hike at 74%.
FAQ
Are prediction markets good at forecasting Fed decisions? On Kalshi, the favourite at the close the day before has been right at all 14 meetings since January 2025. Further out, they are less certain and occasionally wrong.
When did Kalshi's Fed market last get it wrong? Thirty days before the September 2026 meeting, it favoured a hold at 72%; the Fed hiked. No day-before favourite has lost in this period.
What does Kalshi expect today? A hold, at 96% on the close of 27 October 2026.
Where can I get the data? In fed-decision-market-prices-2025-2026.csv, with prices at three horizons, the favourite a month out, volume and Brier scores.
Do Polymarket's Fed markets do as well? We looked only at Kalshi, which has the longest price history. Near a meeting, Polymarket's prices usually agree with Kalshi's within a cent or two.
Data as of the close on 27 October 2026. Sources: Kalshi public market data and historical archive; Federal Reserve statements. Not investment advice.
