Methodology & technical details

The Rating

Each team's rating is solved from sportsbook win-total markets and the full 2026 FBS schedule — it's the number that makes a team's summed game-by-game win probabilities match what the market expects it to win, given who's actually on the schedule. The raw scale has no built-in zero, so everywhere on the site — rankings, team pages, matchups — ratings are displayed as OV Rating: the raw rating minus the average across all 138 FBS teams. A team at 0 is exactly average; above or below tells you which side of average it's on. The offense and defense splits use the same convention against their own averages, and still sum to the team's overall net rating.

Reading a Matchup

Subtracting two ratings (plus ~2.2 for home field) gets you close to a game's spread — but not exactly, and the gap grows for big mismatches. That's intentional: ratings are fit on a blowout-capped scale (real margins clipped at ±28, the standard way to keep score-running from distorting ratings for competitive games), which makes raw rating gaps under-predict real-world margins for lopsided games by up to ~9 points. Displayed spreads apply an empirically fitted, held-out-validated correction for exactly that compression — and it's the corrected number that matches real sportsbook lines. So a 12-point rating gap can honestly display as a 16-point spread; win probabilities use the uncorrected scale they were calibrated on.

The matchup view puts the independent OV Line beside market consensus. Their difference is the market gap. Quantified Factors show historically supported reasons that may explain some of that gap; their rough sum is context, not a replacement betting line. Any remainder stays visible because unidentified information may still matter.

Building My Ratings

My Ratings turns your neutral-field winner choices into a personal ranking. Early picks can move teams substantially because the system has little evidence about your view; additional comparisons anchor them. Your ratings stay separate from the OV Rating and are stored only in your browser.

Using the Ledger

Every Explanatory Factor study belongs in the Ledger—including findings that did not support a separate adjustment. Each entry shows the sample, estimate, validation, limitations, product decision, and available game-level evidence behind what appears on matchup pages.