gen-briefly-explain-win-probability-models-and-one-common-miscon-8be7

Briefly explain win probability models and one common misconception about it.

# Win Probability Models **Win probability models** estimate the likelihood that a team will win a game based on the current game situation (score, time remaining, field position, possession, etc.). They use historical data from thousands of past games to calculate probabilities - for example, an NFL team leading by 3 points with 2 minutes left might have an 85% win probability. ## Common Misconception **"A team with 90% win probability *should* win."** This confuses probability with certainty. A 90% win probability means that teams in similar situations won about 90% of the time historically - but it also means they *lost* 10% of the time. If you watch 10 games where teams have 90% win probability, you should *expect* to see roughly one upset. The model isn't "wrong" when a team with high win probability loses - that's exactly what should happen occasionally. The model would only be poorly calibrated if teams with 90% win probability were winning significantly more or less than 90% of the time across many games.