Prediction methodology
By XWhiz Data Team · Updated 2026-09-15 · Model Dixon-Coles v3
Each match is evaluated as an independent event. Team strength comes from Elo-style ratings blended with real league standings (attack and defence per game vs the league average). Expected goals feed a Dixon-Coles adjusted Poisson model, which produces a full 0–6 goal score matrix.
- 1X2: home/draw/away probabilities summed from the matrix.
- Over/Under 2.5 & BTTS: derived from the same matrix, not separate formulas.
- Correct score: the matrix argmax — the single most probable scoreline — so scores vary naturally.
- Odds shown: fair odds (100 ÷ probability), not bookmaker prices.
What the model cannot see
Late injuries, suspensions, rotation, weather and motivation are not in the numbers. Treat every pick as a probability estimate with stated confidence — never a guarantee of profit. Predictions refresh daily at 06:00 UTC.
Responsible gambling: our predictions are statistical analysis for information and entertainment — they are not a guarantee of profit. 18+. Never bet more than you can afford to lose. · Get responsible gambling help