Follow teams and keep your fantasy leagues on every device.

How the CFB Predictions Work

Our call on each game is a predicted margin (shown as a spread), a total and a win chance, all from the GordStats ratings.

How It Works

Margin: the home team's rating minus the visitor's, plus home field (none at a neutral site). Total: a second model rates how high- or low-scoring each team's games run; both teams' numbers are added to a typical game's total, nudged up in blowouts (a soft floor).

Win chance: on games it hadn't seen, the model typically missed by about 16 points. Allowing for misses that size (a normal distribution with a 15.82-point standard deviation), a 7-point favorite wins about 67% of the time.

The record: each prediction is saved before kickoff and graded on that number. Against the spread (DraftKings, via ESPN), only a lean counts: a game where we differ from the book by 3+ points.

Keep in Mind

Predicting all 4,238 FBS-vs-FBS games of 2020-25 from earlier games only, it picked 72.8% of winners. The closing line (the last line before kickoff) is sharper: its typical miss is 15.4 points to our 15.8, and our leans won 50.9% against the spread, about a coin flip.

It ignores injuries (cards only name them), weather and depth charts.

When It Updates

Four times a day, around 5:30 and 11:30 AM and PM Eastern, with every newly finished game and the latest lines.

For the Curious

During a game, the schedule and each game page also show ESPN's live win probability, labeled as ESPN's; ours stays the number we set before kickoff. For the first five weeks of 2026 every FCS team was priced as one generic opponent, which cost those games about 2 points a game; the record keeps those predictions as they were made. The blowout nudge: where the math leaves the loser under 10 points, the total rises by 0.6 of the shortfall, since losers rarely score that few.

Where to find it: CFB predictions · CFB schedule

Related: CFB rankings · the weekly leans · game previews

← Every explainer