How accurate are Polymarket NFL and college football odds?
Two tables answer it, one per league, from the 2025 season’s game-winner markets on Polymarket. Each market is scored on its last traded price 24 hours before it resolved, which for every game counted fell before noon Eastern on game day. The favorite is whichever team was priced above 50 cents at that moment, and the question each row answers is how often that team went on to win. These are Polymarket’s prices, not any trader’s record. Figures as of 2026-09-21.
NFL, 2025 season
283 of 283 game-winner markets in our record (regular season and playoffs, no preseason; games from September 4, 2025 to February 8, 2026) had a traded price at the 24-hour cutoff.
| Favorite priced at | Games | Stated price (average) | Favorite won | 95% interval on the win rate |
|---|---|---|---|---|
| 50 to 60 cents | 83 | 55.3% | 61.4% | 51-71% |
| 60 to 70 cents | 86 | 64.2% | 58.1% | 48-68% |
| 70 to 80 cents | 70 | 74.6% | 74.3% | 63-83% |
| 80 to 90 cents | 39 | 84.5% | 82.1% | 67-91% |
| 90 to 100 cents (too few games to quote) | 5 | 90.2% | 100.0% | 57-100% |
Each band with at least 30 games has its stated price inside the 95% interval on the favorite’s win rate. The intervals are wide because a season is a few hundred games. The 60 to 70 cent favorites won 58.1% of 86 games, and the interval on that rate runs from 48% to 68%. Favorites priced at 90 cents or more (5 games) are shown in the table but too few to quote.
Scope: Game-winner markets only, one per game, from the 2025 season (regular season and playoffs, no preseason). Each is scored on its last traded price 24 hours before the market resolved, which for the games counted fell before noon Eastern on game day. The favorite is whichever side was priced above 50 cents at that moment. This measures Polymarket prices, not any trader; games with no trade at the cutoff are absent. A band with fewer than 30 games is shown but never quoted.
College football, 2025 season
797 of 831 game-winner markets in our record (games dated between August 1, 2025 and January 31, 2026) had a traded price at the 24-hour cutoff; the rest had no trade at that moment and are absent.
| Favorite priced at | Games | Stated price (average) | Favorite won | 95% interval on the win rate |
|---|---|---|---|---|
| 50 to 60 cents | 160 | 54.7% | 53.8% | 46-61% |
| 60 to 70 cents | 190 | 64.7% | 64.7% | 58-71% |
| 70 to 80 cents | 171 | 74.5% | 73.1% | 66-79% |
| 80 to 90 cents | 147 | 84.2% | 86.4% | 80-91% |
| 90 to 100 cents | 129 | 94.6% | 96.9% | 92-99% |
Every band resolved within 2.3 points of its stated price, and every stated price sits inside the 95% interval on its win rate. With 129 to 190 games per band the intervals are narrower than the NFL’s.
Scope: Game-winner markets only, one per game, from the 2025 season. Each is scored on its last traded price 24 hours before the market resolved, which for the games counted fell before noon Eastern on game day. The favorite is whichever side was priced above 50 cents at that moment. This measures Polymarket prices, not any trader; games with no trade at the cutoff are absent. A band with fewer than 30 games is shown but never quoted.
How to read these tables
A calibrated price is one where favorites priced at 70 cents win about 70% of the time. The stated price is the average price in the band; the win rate is how often the favorite actually won; the interval is the range a season of that many games can pin the true rate to (a 95% Wilson interval). A stated price inside the interval means the season’s results are consistent with the prices; it does not prove the prices were exactly right, and a second season would move every band. Only the game winner is scored here; point spreads and totals are not. The same method, applied across Polymarket’s markets rather than one sport, is the calibration benchmark; how wallets rather than prices are graded is on the methodology page.
How often does the NFL favorite win on Polymarket?
For 2025-season games, with the favorite taken from the price a day before the market resolved: favorites at 50 to 60 cents won 61.4% of 83 games (95% interval 51-71%), at 60 to 70 cents 58.1% of 86 (48-68%), at 70 to 80 cents 74.3% of 70 (63-83%), and at 80 to 90 cents 82.1% of 39 (67-91%). Each band's stated price sits inside the 95% interval on its win rate. Favorites priced at 90 cents or more (5 games) are shown in the table but too few to quote. The favorite is whichever team was priced above 50 cents at that moment, and this measures Polymarket prices, not any trader.
How often does the college football favorite win on Polymarket?
For 2025-season games, with the favorite taken from the price a day before the market resolved: favorites at 50 to 60 cents won 53.8% of 160 games (95% interval 46-61%), at 60 to 70 cents 64.7% of 190 (58-71%), at 70 to 80 cents 73.1% of 171 (66-79%), at 80 to 90 cents 86.4% of 147 (80-91%), and at 90 to 100 cents 96.9% of 129 (92-99%). Every band resolved within 2.3 points of its stated price. The favorite is whichever team was priced above 50 cents at that moment, and this measures Polymarket prices, not any trader.
Are Polymarket NFL odds accurate?
For the 2025 NFL season's game-winner markets, priced a day before they resolved, every price band with at least 30 games had its stated price inside the 95% interval on the favorite's actual win rate. That is a statement about the prices over the 278 games in the four bands with at least 30 games each, not about any trader; a single season of 283 games cannot pin the true rates closer than those intervals.
The sports wallets, graded against what they paid
This page scores the prices. The sports board scores the wallets trading these games, against the prices they paid, and its #1 forecaster is free to read with the full record.
See the sports board’s #1 forecaster →To grade one wallet yourself, paste it into the free checker.
Game-winner markets only, one per game, from the 2025 season, scored on the last traded price 24 hours before resolution; outcomes spot-checked against Polymarket’s public API. A slice of the calibration benchmark’s method (Polymarket’s CTF Exchange and NegRisk markets; Up or Down series are excluded by policy). Figures as of 2026-09-21, computed on the pipeline data that serves the product and cross-checked before publication. Nothing here is financial or investment advice. Past accuracy does not predict future outcomes.
The Weekly Skill Report. Leaderboard changes, notable wallet moves, and every newly graded alert from the featured feed, the alerts subscribers receive. Wins and losses both, one email each Monday.
Free. Unsubscribe anytime. Privacy.