How our predictions work

Team ratings

Every team gets attack and defense ratings fitted on multi-season results, with recent matches weighted more heavily (a match a year old counts about a quarter of one played last week). Ratings are fitted per country, so cup games connect divisions and promoted teams inherit sensible starting points.

From ratings to probabilities

For each fixture the ratings produce expected goals for both sides, expanded into a full scoreline probability grid with a correction for the draw-heavy low-scoring outcomes football is known for. Match result, correct score, goals over/under, both-teams-to-score, double chance and Asian handicap probabilities are all read from that one grid.

A dedicated corners model

Alongside goals, corners get their own ratings: how many corners a team wins and concedes, fitted the same way. The model projects corners for each side and prices the actual corner line of the match — one of seven markets covered for every fixture.

Blending with the market

Where bookmaker lines exist, model probabilities are blended toward market-implied ones — heavily for teams with thin history, lightly for well-covered leagues. When a value gap of 5% or more remains after blending, the pick is flagged as value.

No hand-picking

Every eligible real-football match gets a prediction — no cherry-picking the easy ones. Esports and virtual leagues are excluded entirely. Predictions freeze 30 minutes before kickoff, and every frozen pick is settled against the final result and counted in the public accuracy record, wins and losses alike.