How BEEF works

BEEF rates on-field team performance in the trenches. The offensive index minus the defensive index gives a signed BEEF score: positive means above the national baseline, and negative means below it.

Reading the score

The offensive and defensive indices share a baseline of 100. Higher offense is better; lower defense is better. An offense of 115 and defense of 90 produce BEEF +25.0. Improving either side raises BEEF, and weakening either side lowers it.

The indices combine several measures on a common scale. They are not points, yards, win probabilities, or predicted scoring margins. There is no final 0–100 cap.

Inputs

Offense uses line yards, stuff avoidance, power success, rushing success, front-seven havoc avoidance, PPA, and explosiveness. Defense uses the opposing versions of those measures.

Adjustment

Game-level results are adjusted for opponent strength, normalized across FBS teams, and tempered when a team has played few qualifying games. FBS-versus-FCS games appear in records but do not enter BEEF scores.

The existing weighted offensive and defensive strength composites are standardized using their pooled national mean and standard deviation. Offense is 100 plus ten times its standardized strength; defense is 100 minus ten times its standardized strength. Both sides have equal influence. Scores are calculated before display rounding, and teams are ranked by the unrounded difference.

Team profiles

The five season comparisons show rushing and blocking production through the selected snapshot, including FCS opponents. These descriptive stats are not opponent adjusted. Missing inputs remain unavailable, and partial coverage is identified where needed.

Sources and history

Ratings use CollegeFootballData game-level advanced statistics and a separate havoc feed. Published snapshots are versioned; a corrected source creates a new version rather than silently changing the archived calculation. Missing values are excluded, never treated as zero.

Signed ratings use beef-v0.5.0. Older archived versions use the legacy 0–100 scale, where both offensive and defensive ratings are higher-is-better. Seasons without havoc use a separately labeled variant that rescales the remaining weights. Scores are relative to their own snapshot; they are not calibrated for direct comparisons across seasons or variants.

Polls, recruiting, player size, and brand do not enter the score.