WORLD CUP2026Match Predictor

1. 01

   Team strength (Elo)

   Every nation carries an Elo rating fitted from years of historical international results. Stronger results raise a team's rating; the gap between two ratings is the single input that drives every prediction. Finished World Cup matches feed back in and nudge the ratings as the tournament unfolds.

2. 02

   Match model (Dixon–Coles Poisson)

   The Elo gap is converted into each side's expected goals. We then build the full grid of possible scorelines with a Dixon–Coles Poisson model (a low-score correction makes 0-0, 1-0 and 1-1 realistic) and sum it into win / draw / loss probabilities and the most likely exact scores.

3. 03

   Tournament simulation (Monte Carlo)

   To get title and stage odds we replay the whole tournament 20,000 times. Each run simulates all 12 groups, ranks them, fills the knockout bracket (including the eight best third-placed teams), and plays every match through to the final. Counting outcomes across all runs gives each team's chance of reaching each round.

4. 04

   Data & freshness

   Fixtures, live status and final scores come from the football-data.org feed. The whole pipeline re-runs on a schedule, so probabilities refresh as real results land. Figures are model estimates for entertainment — not betting advice.

Simplifying assumptions: group and knockout matches are treated as neutral-venue; the best-third → bracket-slot mapping uses a greedy assignment rather than FIFA's full 495-scenario table; deeper-round pairings follow the published bracket order.