How PitchModel prices a football match
A plain-English look at the Poisson / Dixon-Coles model behind every probability on the board.
PitchModel does not pick winners from gut feel. Each fixture is priced from league strength tables, recent form and a Poisson score grid — the same family of models used in academic football analytics.
Attack and defence ratings are estimated for both teams, adjusted for home advantage measured from real results, then turned into a full matrix of scorelines. From that matrix we read match odds, over/under, BTTS and related markets.
Early in a season the current table is thin, so last-season data is blended in so sides stay differentiated instead of collapsing to the same number. When team news is available, availability is applied as a light multiplier on attack or defence.
Every public pick is logged and graded on the track record page. That is the feedback loop: the model is only as useful as the results it leaves behind.