Season grade leaderboard

College grades use the same method as the NFL: every play's expected points added, adjusted for how good the defense was, ranked against every other player at the position. The play data comes from CollegeFootballData.com, with ESPN's play feed saying who threw, ran and caught the ball. Four things are different.

• Only games against FBS teams count. A power-conference offense against an FCS defense tells us little about either, so those games show in the game log but aren't graded.
• Garbage time is left out: plays with a lead of more than 38 points in the 2nd quarter, 28 in the 3rd, or 22 in the 4th. Blowouts are common in college, and backups piling up numbers against backups would otherwise count like starters.
• No win probability or completion-percentage data. The college play feed doesn't have them, so grades rest on opponent-adjusted EPA and success rate (the share of plays with positive EPA).
• Grades and leaderboards use a minimum workload: every grade is a percentile among the players who meet it, so a few plays can't skew anyone else's. Per game against FBS teams, 14 dropbacks for quarterbacks, 10 touches for running backs, 5 targets for receivers and 3 for tight ends, counted over at least two games. Defenders need to have played in half their team's games.

Defensive grades use the same box-score impact score as the NFL's, minus forced fumbles, which the college box score doesn't record.