How to read win probability: what a 60% chance really means
What a win-probability percentage is, why 60% is not a guarantee, how it differs from odds, why the three numbers add to 100%, and how to judge whether a model is any good.
What the percentage means
A win probability is a model's estimate of how often a result would occur if the match were played many times over. If a team is given a 60% chance to win, the claim is that in 100 replays of this fixture, from this starting point, it wins about 60 of them. It is a statement about likelihood, not certainty — which is exactly why the other 40% still happens often enough to matter.
Why 60% is not a promise
The single most common mistake is reading 60% as “they will win.” They will not, four times in ten. A favourite losing is not evidence the number was wrong — a fair coin still lands tails half the time. You can only judge probabilities across many predictions: if everything a model calls 60% wins roughly 60% of the time, the model is well-calibrated, however any single game turns out.
Probability vs odds
Bookmakers quote odds; models quote probability, and the two are two sides of the same coin. Decimal odds of 2.00 imply a 50% chance (1 divided by 2.00); odds of 1.50 imply about 67%. The gap worth knowing is the margin: published odds are shifted so the implied percentages add up to more than 100%, building in the bookmaker's edge. A clean probability model has no such thumb on the scale.
Why the numbers add to 100%
In football a fixture has three outcomes — home win, draw, away win — so a proper forecast gives three probabilities that sum to 100%. That draw slice is why football win probabilities look lower than in sports without draws: a 45% home / 27% draw / 28% away split still makes the home side clear favourites. In basketball, with no draw, the two numbers simply split 100% between the sides.
How to tell a good model from a guess
Do not judge a forecast by whether the favourite won; judge it over time by calibration and by whether it beats simple baselines. A trustworthy model is transparent about the inputs behind the number — form, home advantage, team strength, schedule — rather than presenting a percentage as if handed down. On SportMetrics the probability is the headline, but the breakdown beneath it is the point: the reasoning is what you should weigh, not the single figure.