Inside the Numbers: How Accurate Are College Football Betting Lines?

Analysis · by Rick's Picks Analytics

How good is Vegas at setting college football lines? We broke down every FBS game from 2015-2024 to measure spread accuracy, over/under efficiency, and identify the specific conditions where oddsmakers struggle.

Spread Accuracy by the Numbers

For every game in our dataset, we compute the spread error -- the difference between the predicted margin (the spread) and the actual margin. A perfect line means the spread error is zero.

What the distribution looks like: The median spread error is close to zero, confirming that Vegas is centered correctly on average. But the distribution is wide -- roughly 30-40% of games finish within 3 points of the spread, while about 10-15% of games finish more than 14 points away from the spread. College football is inherently volatile, and even the sharpest oddsmakers cannot predict blowouts versus nail-biters.

Season-by-Season Trends

We track home cover rates and over/under rates for each season from 2015 through 2024. This reveals whether the market has structural biases that persist across years.

The verdict: Neither home cover rates nor over rates show a persistent multi-year trend. They oscillate around 50%, sometimes favoring one side for 2-3 consecutive seasons before reverting. This is consistent with an efficient market that occasionally displays short-term noise.

Weather and Totals

This is where it gets interesting. We segment games by weather conditions and test whether cold, warm, and dome environments affect over/under rates.

Cold weather (below 40F): Games played in freezing conditions show a measurable shift in total scoring. Our binomial tests compare cold-weather over rates against baseline, with Cohen's d measuring practical significance. Cold tends to suppress scoring, pushing results toward the under -- but only when the temperature drops below 40 degrees. Cool weather (40-55F) shows minimal effect.

Dome games: Games played in controlled environments (domes) strip away all weather variables. We test whether dome over rates differ from outdoor games and find that domes tend to produce slightly higher-scoring affairs, consistent with perfect conditions favoring offensive execution.

Warm weather: Games above 70F don't show the dramatic effects that cold weather does. Warm-weather scoring is close to the overall average.

How we use it: When our algorithm projects totals for a game, we apply a weather adjustment factor. Below-freezing games get a slight under bias, dome games get a slight over bias, and moderate temperatures use no adjustment. This is one of several "edge-stacking" inputs.

Conference ATS Performance

Do certain conferences consistently beat or underperform against the spread? We compute per-conference cover rates for each Power 5 conference, carefully avoiding the double-counting bug that plagued our earlier analysis (where games between two teams in the same conference were counted twice).

The finding: Conference-level ATS differences exist but are generally small (within 2-3 percentage points of 50%). No single conference provides a reliable, persistent betting edge. The differences that do appear tend to rotate year-to-year rather than reflecting a structural advantage.

Takeaway for Our Prediction System

Vegas is very good at its job. The exploitable edges are not in broad market-level biases but in game-specific factors: weather, public perception of ranked teams, and home-field overvaluation. Our system stacks these small advantages together, each contributing a fraction of a point, to produce a composite prediction that outperforms any single factor.

All tests use Bonferroni correction across the full family of comparisons, with separate validation on 2023-2024 holdout data.

Rick's Picks publishes statistical analysis of college football for informational and entertainment purposes. Nothing here is betting advice. 21+. If gambling is affecting you or someone you know, call 1-800-GAMBLER.