How Sportiko Predicts Football Matches
Every prediction is a probability, not a promise. This page explains what goes into the model, how to read its output, and where its limits are.
What feeds the model
For each match, an expected-goals estimate is built for both teams from several independent signals, weighted and combined:
- Recent team form, adjusted for the quality of the opposition faced
- League standings and points-per-game strength ratings (FIFA rankings for national teams)
- Head-to-head history between the two sides
- Home and away records
- Squad availability and injuries
- Bookmaker market consensus, used as a calibration anchor
- Per-team scoring patterns, including when in the match each team tends to score and concede
How to read the probabilities
The expected goals feed a Poisson-based score matrix, which produces probabilities for the match result and every goal market. A 70% probability means roughly 7 of 10 such situations go that way — and 3 do not. Even strong favourites lose; a well-calibrated 70% is the honest claim, not a guarantee.
When predictions update
Predictions are generated as soon as fixtures are confirmed, refreshed daily, and regenerated when significant new information lands — fresh bookmaker odds, lineups or team news. Each prediction shows its last-updated time. During live matches, in-play probabilities update as events happen.
How results are graded
Every prediction is graded automatically against the official full-time result — wins and losses alike — and the aggregate record is published on the track record page. Postponed or abandoned matches are not graded; their predictions are regenerated when the fixture is rescheduled.
Limitations
Football is low-scoring and high-variance: red cards, deflections and individual moments decide matches no model can foresee. Smaller leagues have thinner data and wider uncertainty. Sportiko is a statistics and analytics service for informational purposes — it is not betting advice, and no outcome is ever guaranteed.