College Football Ratings — Week 1

• These ratings are computed here from final scores — they are not a republished third-party rating. Nothing on this page comes from a subscription or permission-gated system.
• SRS is a ridge-regularized least-squares fit: each team's rating is chosen so that (home rating - away rating + home-field) best predicts the margin of every game played. It is a whole-season simultaneous solve, so a late upset re-weighs the entire season.
• Elo walks the games in date order and updates after each one, with diminishing credit for blowouts. It reacts to a team improving mid-season in a way a season-long average cannot, and it carries a regressed rating in from last season.
• Ratings are points relative to an average FBS team on a neutral field, so a difference between two teams is a projected spread.
• Home-field advantage is ESTIMATED from the season's own games rather than assumed. Measured home margin has ranged from about 2 points (2020, no crowds) to nearly 8 by season, so a fixed constant would be wrong in both directions.
• Strength of schedule is the average rating of the opponents a team actually played. Teams below Division I are pooled into a single opponent, so scheduling one cannot inflate a schedule rating.
• Blowouts are capped before they enter the fit, so a 63-0 win counts as a decisive win rather than four times a 17-0 win.
• Teams with fewer than four completed games are not rated: with a shorter schedule the rating reflects the prior more than the season.
• Validated walk-forward on 3,867 held-out games (2018-2025), fitting only on games played before each week: SRS predicted game margin to 12.7 points mean absolute error, against 13.5 for the same model with the schedule adjustment removed and 16.1 for assuming the home team wins.

• These ratings see final scores only — no injuries, no weather, no personnel changes, no market information. A team missing its starting quarterback is rated exactly as if it were not.
• The consensus betting market predicted the same held-out games better (12.1 vs 12.7 mean absolute error). Projected margins here are context, not betting advice, and should not be read as an edge.
• Early-season ratings are unstable by construction. Through the first month a rating rests on three or four results and will move sharply; treat November's ratings as far more settled than September's.
• FCS teams are rated internally because FBS teams play them, but are not listed. A rating is only comparable within the population it was anchored on.

FACT (final scores, schedule, site, conference affiliation), DERIVED (SRS, Elo, strength of schedule, projected margins)