Matchup Predictor
Pick any two FBS teams. The ML engine forecasts the game from each team's play-by-play stats, Elo and roster talent; move a stat to see how the forecast responds.
What's driving it
Points each part of the model adds for one side, from the two teams' stats.
Adjust the stats
Each slider starts at the team's real number (rank among FBS teams beside it). This is the model's sensitivity, not a claim that changing one stat would change the result.
How the ML engine works
A linear model trained on 4,777 FBS games (2016–2025, from CFBD's Model Training Pack): every input is the gap between the two teams in one pregame stat, each worth a fixed number of points, plus home field. Every stat can only push in the direction you'd expect, so a better offense never lowers a team's odds. The margin becomes a win probability through a normal curve fitted on past games.
Tested on 4,468 games, each season scored by a model that never saw it: it misses the final margin by 12.59 points on average, against 13.11 for the power rating behind our picks and 12.26 for the Vegas closing line. On 2025, held out until the end: 12.01 vs 12.43 (490 games).
It is separate from our ranking: the résumé ranking (wins above bubble) counts only results; this engine uses Elo and talent, which carry over from earlier seasons. Stats update Sunday nights; a team on a bye keeps its previous week's numbers.