Coaches on 4th down
Every go, field-goal and punt call since 2018, judged by the 4th-down model: how much win probability a coach gave up per decision compared with other coaches in the same situations.
Aggressiveness vs. decision quality
Each dot is a rated coach. Right = went for it more often; up = gave up less win probability than other coaches in the same situations. Dashed lines: the league.
Leaderboard
Click a column to sort. Lower WP lost is better. Rated: 100+ decisions for the combined view, 50+ in a single season.
How these are computed
- WP lost: our 4th-down model’s win probability for its best option minus the coach’s choice, in WP points (0 when the coach agreed). The model is the registered decision layer with all 101 bootstrap replicates; “confident” means at least 83% of them agree.
- Same situations: each decision is compared with what other coaches lost in the same field position × distance × score × clock cell, plus the season’s league level, so a coach isn’t penalised for facing harder 4th downs.
- Flags (combined view): coaches with 100+ decisions are tested with a bootstrap over their games and the Benjamini–Hochberg procedure at a 50% false discovery rate, so up to half the flags may be false alarms. “–” means not separable from average.
- Everything is relative to our model, which is in-sample for 2018–2025. 2026 is through week 9 and is updated weekly.
- The head coach is credited with every call; coordinators aren’t recorded.