NFL Power Rankings 2024
Pre-season ratings, carried over from last season and regressed toward the league average. They update as soon as Week 1 is played.
Power Rankings
Season progression
Loading season progression…
How these rankings work
Every team carries two ratings, both measured in points against a league-average opponent: an attack rating for the points it scores above average, and a defense rating for the points it prevents below it. Overall is the two added together, so a team rated +7 would be expected to beat an average team by about a touchdown on neutral ground.
After each game, both ratings move on their own error — what the team actually did against what the model expected, given the opponent's ratings and home advantage. Beating a good defense counts for more than beating a poor one, which is why these rankings and the standings disagree: a win-loss record can't tell the difference between a hard schedule and an easy one.
Margins are damped, so a 45-point win says more than a 20-point win without saying twice as much, and ratings carry into a new season at half weight — last year is evidence, not gospel, once a roster has turned over. Win probabilities come from the rating gap and the historical spread of NFL results, rather than from simulated games.
How accurate is it?
The model is replayed week by week over past seasons, predicting each week using only what it knew beforehand. Measured over 272 games in 2025, against the obvious benchmark of always backing the home team:
| Measure | This model | Always home |
|---|---|---|
| Winner picked correctly | 61.9% | 53.9% |
| Average margin error | 10.2 pts | 11.0 pts |
| Brier score (lower is better) | 0.219 | 0.248 |
Brier score measures how well-calibrated the probabilities are, not just whether the pick was right — 0 is perfect, 0.25 is a coin flip. Figures are recalculated every week across every completed game, so they include the season in progress.
