Methods
Calibration
Whether forecast probabilities match how often things happen: of all the teams given a 70% chance, about 70% should win.
How we use it
Our track record groups completed games by the favorite's win probability and compares the average forecast with how often the favorite actually won. Big gaps in either direction mean the model is over- or under-confident.
Where to see it
Related terms
- Brier Score
- A measure of probability forecasts: the average squared gap between the forecast and what happened (1 if it happened, 0 if not). Lower is better; always saying 50% scores 0.25.
- Win Probability
- The model's estimated chance that a team wins a game, from the share of simulated games it wins. A 70% favorite should still lose about 3 games in 10.