FOOTPredictions › Correct score probabilities
Probabilities, graded in public
Correct score probabilities
How a model assigns a probability to every scoreline, why the top pick is usually under 15%, and today's fixtures to try it on.
What it means
A correct score probability is the model's estimate for one exact final score — 1–1, 2–0, 2–1 and so on. Football is low-scoring, which keeps even the likeliest lines modest: in a balanced match, 1–1 at 11–12% is a typical leader.
Scorelines are where the model is at its most transparent: instead of one number, you see the whole distribution — and Footalytics publishes the top lines for every official prediction, then grades them against the real score.
How the model prices it
Expected goals in, scoreline matrix out: the Dixon-Coles Poisson model computes a probability for every plausible score from the two sides' goal expectations, correcting the raw model's known bias toward too few 0–0 and 1–1 draws. The "most likely score" is simply the matrix's largest cell.
Official predictions only, settled against real final scores — see the dashboard for windows and league splits.
Today's matches
One tap asks the AI this exact question for that match.
What the numbers are worth
Because dozens of plausible scorelines share the probability mass, exact scores are the hardest thing to call in football — a 12% favourite is still more likely to lose than to land. That honesty is the point: use the top-three lines to understand the shape of the match, not as a single expected outcome.
FAQ
What is a correct score prediction?
An estimate of the exact final score. The model produces a full distribution — every scoreline gets a probability — and we publish and grade the top lines for every official prediction.
Why is the top scoreline's probability so low?
Probability mass is spread over dozens of plausible scores in a low-scoring sport. In a close match even the leader (often 1–1 or 1–0) sits around 10–13%. Low numbers are honesty, not weakness.
Which scorelines does Footalytics publish?
The top three most likely scorelines with their probabilities on every official prediction, plus the exact-score hit rate on the public accuracy dashboard.
Can I ask the AI about a scoreline in my own words?
Yes — ask things like "most likely score in Arsenal vs Chelsea" and the assistant answers with the model's distribution and the reasoning behind it.