How accurate is GPT-5.6 at football predictions?

openai

Real predictions, graded against real results — a live experiment, not a proven edge. Last 30 days, recomputed after every graded match (updated 1 Sept 2026, 21:44 UTC).

Result accuracy
51.7%
Exact score
10.1%
Calibration
53.5%
Graded predictions
592
01 — Confidence & calibration

Across 510 predictions with a stated confidence, GPT-5.6 averaged 49.7% confidence — 53.2% when it was right vs 46.2% when it was wrong. Calibration score: 53.5%.

02 — By competition (30 days)
CompetitionGradedResult acc.Exact score
Major League SoccerUSA6142.6%16.4%
UEFA Europa LeagueWorld4866.7%10.4%
ChampionshipEngland4438.6%4.5%
Serie ABrazil4040.0%7.5%
Pro LeagueSaudi-Arabia3554.3%8.6%
EredivisieNetherlands3360.6%9.1%
DFB PokalGermany3180.6%19.4%
La LigaSpain3060.0%16.7%
AllsvenskanSweden2941.4%3.4%
2. BundesligaGermany2744.4%7.4%
Liga MXMexico2751.9%3.7%
Süper LigTurkey2740.7%7.4%
UEFA Champions LeagueWorld2653.8%3.8%
Premier LeagueEngland2055.0%15.0%
Serie AItaly2070.0%20.0%
BundesligaAustria1968.4%10.5%
Ligue 1France1838.9%5.6%
EliteserienNorway1838.9%11.1%
CONMEBOL LibertadoresWorld1631.3%12.5%
Super LeagueSwitzerland1450.0%7.1%
BundesligaGermanylow sample966.7%11.1%

Low-sample rows are dimmed, not hidden — transparency beats tidiness.

03 — Pick value (30 days)

Every pick scored against the real market price recorded at prediction time, tracked in units — a live experiment, not betting advice. 18+. Please play responsibly.

Win rate
42.0%
Pick value
-8.3%
Units
-49.1u
Avg market price
2.50
Graded
590

← Full model leaderboard · how grading works · which AI is best? · compare prediction tools