Are Vegas Spreads Actually Beatable? What 10 Years of Data Tells Us

Analysis · by Rick's Picks Analytics

Every college football fan has a theory about beating the spread. "Home teams always cover." "Vegas overvalues ranked teams." "Big favorites never cover." We put these to the test with rigorous statistical analysis across 10 seasons of FBS data.

Our Methodology

We applied five formal hypotheses to every FBS game from 2015 through 2024 using a temporal train/test split -- training our models on 2015-2022 data, then validating on the unseen 2023-2024 seasons. Every test uses Welch's t-test for means, binomial tests against the 52.38% breakeven rate (the threshold needed to profit at standard -110 odds), Cohen's d for effect size, and Bonferroni correction to guard against false positives from running multiple tests.

Hypothesis 1: Are Vegas Lines Efficient?

The baseline question: does the home team cover exactly 50% of the time, as efficient-market theory predicts?

Our analysis tests three splits -- full dataset, training set, and test set -- using binomial tests against both 50% (pure efficiency) and 52.38% (the breakeven needed to actually profit at -110 juice). We also compute the implied ROI: if you blindly bet every home team against the spread at -110 odds, would you make money?

What we found: Home cover rates hover remarkably close to 50%, generally within 1-2 percentage points. The ROI on blindly betting home teams is slightly negative after accounting for the vig. Vegas is efficient in aggregate -- but that doesn't mean there aren't pockets of inefficiency in specific situations.

Hypothesis 2: The Home Favorite Bias

This is one of the most persistent theories in college football: the public loves betting home favorites, which inflates their lines and makes them less likely to cover. We test whether home favorites cover at a statistically lower rate than road favorites using chi-square tests and Cohen's d for effect size.

The theory: When a popular home team like Alabama hosts a mid-tier opponent, casual bettors pile on. The sportsbook adjusts the line to balance action, pushing the spread beyond what the data supports. The result: home favorites cover at a lower rate.

What the data shows: There is a measurable difference between home favorite and road favorite cover rates. Home favorites tend to cover slightly less often than road favorites, consistent with the public-money theory. However, the effect size (Cohen's d) is typically small -- meaning the bias exists but is modest. Whether it survives Bonferroni correction depends on sample size and the specific season window.

How we use it: In our prediction algorithm, when a popular home team is favored by a large spread, we apply a slight contrarian adjustment. The data supports fading heavily-bet home favorites, particularly when the spread exceeds two touchdowns.

Hypothesis 3: The Over/Under Trend

College football scoring has been rising for years. Has the over been hitting more often as a result, or do oddsmakers adjust?

We track over rates season-by-season from 2015-2024 and test whether the over hits above the 52.38% breakeven threshold using binomial tests.

What we found: Over rates fluctuate significantly by season, but across the full 10-year window, they hover near 50%. Oddsmakers are generally good at adjusting totals upward to account for rising scoring trends. There is no persistent, exploitable over bias in the aggregate data.

Hypothesis 4: Large Spread Accuracy

Do blowout-sized spreads (greater than 14 points) behave differently than close spreads (7 points or fewer)?

Why it matters: Large spreads represent games where one team is heavily favored -- think Ohio State hosting a MAC team. The question is whether Vegas is as accurate in these mismatches as in competitive games.

We compare favorite cover rates, ATS margins, and apply chi-square tests across spread tiers. Cohen's d measures whether the practical difference is meaningful.

What the data shows: Large favorites do cover at a different rate than small favorites, but the direction varies across seasons. The key finding is that ATS margins in blowout-spread games have higher variance -- Vegas has a harder time pinpointing the exact margin in mismatches, which creates potential value on underdogs in these spots.

Hypothesis 5: Ranked Team Public Bias

Our most nuanced test: do ranked teams (Top 25) cover less often because the public overvalues the prestige of a ranking?

We use a vectorized, long-form analysis where every team appearance in every game gets its own row, allowing us to separately analyze ranked favorites, ranked underdogs, unranked favorites, and unranked underdogs.

Important caveat: Our ranking data reflects current-season rankings, not the exact game-time ranking. This introduces some look-ahead bias, so we treat these results as exploratory rather than confirmatory.

The finding: Ranked teams, particularly ranked favorites, show a slight tendency to underperform against the spread. This aligns with the "public overvaluation" theory -- casual bettors see a ranking and assume the team is a lock. The effect is most pronounced for ranked home favorites with large spreads, where multiple public biases compound.

The Bottom Line

Vegas is remarkably efficient in the aggregate. You cannot profit by blindly betting one side. But specific situations -- heavily-bet home favorites, large spreads, and ranked teams with inflated public perception -- show measurable biases that our algorithm exploits. The key is combining multiple small edges rather than relying on any single angle.

This analysis uses Bonferroni-corrected significance thresholds to avoid false discoveries, and validates all training-set findings on held-out 2023-2024 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.