How it works

Elo is a rating system that estimates how strong a football team is relative to every other team.

After each match, the winner gains points and the loser loses some; a bigger upset shifts ratings more than a predictable result.

Signal uses these ratings as a core input to compare teams before every fixture.

The model estimates expected goals (λ) for each side, using Elo, recent form, and contextual features.

Those λ values build a score matrix via independent Poisson distributions for home and away goals.

Summing the matrix cells yields match probabilities: home win, draw, away win, and exact scores.

Every week, Signal scores its predictions on finished matches using log-loss - a metric that penalizes overconfident wrong calls.

We publish the full history and compare directly against bookmaker odds, the hardest realistic benchmark.

Track the numbers week by week on the Performance page - no hidden weeks, no cherry-picked results.