Definition
Brier Score measures how accurate a probability forecast was after the result is known. For a single yes/no bet:
Brier Score = (forecast probability - outcome)^2
The outcome is 1 if the event happened and 0 if it did not. Lower is better. A perfect forecast scores 0; a very bad confident forecast can approach 1.
Worked Example
A team is listed at -110, which implies a breakeven probability of:
110 / (110 + 100) = 52.38%
Your model gives the team a 55% win probability.
If the team wins:
(0.55 - 1)^2 = 0.2025
If the team loses:
(0.55 - 0)^2 = 0.3025
Across many bets, average the individual scores. If one model averages 0.215 and another averages 0.240 on the same set of games, the 0.215 model made better probability forecasts.
Why It Matters
Brier Score helps a bettor test whether model probabilities are calibrated, not just whether picks won or lost. It is useful when comparing betting models, tracking forecast quality, and deciding whether an apparent edge over a sportsbook line is backed by accurate probabilities.