luckadjusted

Every game, adjusted for luck.

The scoreboard tells you who won. It doesn’t tell you how much of that was the bounces. So we replay every game 4,000 times, using the plays both teams actually ran, and count how often each side comes out on top. It’s a model, not a ruling. Everything that goes into it is on the methodology page.

Soccer

Every match replayed 4,000 times. The current Premier League, La Liga and Bundesliga seasons and the 2026 World Cup are replayed from both teams’ goals (public-domain results data has no shots), earlier tournaments and seasons from every shot and its expected goals (StatsBomb’s open data): lucky or expected, who deserved to win, and who got lucky.

How it works

A final score is part performance, part bounces. To pull the two apart, we take each team’s plays from a game, shuffle them into new drives and play the game again. 4,000 times. Kicks, fumbles and flags land at their normal rates, not where they actually landed.

If a team wins 70% of those replays, that’s roughly what its performance was worth. Whatever’s left between that and the real result, we call luck.

Does it hold up? On real rematches it picks the winner about as well as the betting line. Every test is public: the NFL method, what’s different for college and for hockey.