Football analysis

Behind every FootIQ prediction is a match analysis: the numbers the models produced and the context the AI reviewed. This page explains what each part means.

Expected goals per model

Each of the three models (Dixon–Coles, bivariate Poisson and empirical Bayes) estimates how many goals each team will score. When the models agree, the prediction is more robust; large disagreement is a warning sign and can stop a pick from being published.

Market probabilities

From the expected goals the engine derives a full probability table of possible scores and, from it, each market's probability. Probabilities are calibrated on past out-of-sample predictions and blended with the bookmaker's view where a price exists.

Form and venue

Recent results are split into home and away form. Matches further in the past weigh less (the weight halves after about a year).

xG and chance quality

Expected goals (xG) measure the quality of chances a team created, not just the goals it scored. See understanding xG.

Head-to-head, standings and absences

Previous meetings, league position and reported injuries or suspensions are shown for context. When the starting lineups are announced before kick-off they are included too.

Data quality

Every analysis carries a data-quality measure based on how much reliable history is available for both teams. Matches with too little data are not published, however attractive the numbers look.