Explore win, draw and loss probabilities from our statistical model, with separate AI forecasts for selected leagues. Predictions are locked before kickoff and checked against the final results. Wins and misses stay on record.
Import verified sourcesFixtures, results and team news from permitted sources. Results are cross-checked against a second source when one is available.
Calculate probabilitiesElo and Dixon-Coles ratings turn form into 90-minute probabilities.
AI analysts reviewIn selected leagues, Claude and Codex explain the numbers; their own views run in shadow.
Lock and gradePicks lock before kickoff (normally 24 hours ahead), then get graded in public.
Yes, everything is free and there is no account. Every probability and every market is on the page, and so is the AI analysts’ reasoning where they run.
A statistical model (Elo plus Dixon-Coles) produces the published probabilities. Claude and Codex read sourced team news, explain the numbers and give their own view, which we grade separately.
Predictions lock before kickoff, normally 24 hours ahead. We publish a SHA-256 hash of each locked prediction in a public ledger, so anyone can check it was not changed after the match.
Football is hard to predict: even strong models get the 90-minute result right only a little over half the time. We publish every graded pick, misses included, with RPS and calibration on the track record page. Numbers are marked as an early sample until we have 6–8 weeks of live results.
Can we count your visit?With your permission we use Google Analytics, which sets cookies, to see which pages people read. No ads. The site works the same if you say no.
Monthly reportOctober 2026: our first graded forecastsThe six forecasts graded on 31 October, the first day of our public record: how the statistical model did against a simple benchmark, and the results it did not expect.AI reconstruction · Camp Nou, pre-2023