The “Prove It” Rankings

Brawlers (FBS Top 50): For each game decided by fewer than 28 points against an above-median FBS opponent, recalculate the entire model without that game. Keep positive rating lifts and weight them by opponent strength: zero at the FBS median, rising to one at the highest FBS rating. Sum those weighted lifts. Strong competitive performances count; simply losing to a good team does not. The number is an index, not additive rating points or a weekly rank change.

Cupcake Annihilators (all FBS): In each win over a lower-rated opponent, multiply the rating gap by the fraction of maximum margin credit earned. A 28-point win receives full credit; smaller wins receive proportionally less using the model's winner bonus. Average across all games, including losses and wins over stronger teams as zero. Higher scores mean more dominance against weaker opposition, not proof that a team is overrated.

Playoff picture: Our ratings choose the field and project each round. We play out the remaining conference schedule at current strength, keeping results already on the field, then project neutral-field title games between each Power Four conference’s top two teams. Standings ties use head-to-head for two tied teams, then model rating; this simplifies official conference tiebreakers. The ACC, Big Ten, Big 12 and SEC champions qualify, plus the highest-rated team from the American, C-USA, MAC, Mountain West, Pac-12 or Sun Belt—even without a conference title. Notre Dame is guaranteed a place if rated in the top 12. Remaining spots go by rating; seeds 1–4 get byes, seeds 5–8 host, and later rounds are neutral. This is our model’s bracket, not a prediction of the committee’s choices. 2026 CFP rules · Weekly bracket details.

Both use the selected week's ratings and include the existing prior. These are exploratory descriptions of results, not tested improvements to predictions. Download the scores and game-by-game calculations.