The Numbers
The things college bettors argue about, measured on our own archive of 8,440 FBS games with a closing line and a final score. Sample size and a confidence range travel with every number, because “big favorites cover 48%” means nothing without them.
Do big favorites cover?
The question only college football can ask — no other league prices 40-point spreads. The answer is no, and also not the opposite: laying a huge number is a coin flip, exactly like laying a small one.
Bar = 95% confidence range, tick = measured rate. The faint line is 50% (pure chance); the green line is 52.38%, the break-even rate at −110 once the hold is included. Bars falling short of it describe historical rates below break-even.
Every band sits inside its own margin of error. The folk belief that huge college favorites fail to cover is not supported — 28–35 point favorites covered 46.6% and 35+ point favorites 48.1%, both close enough to a coin flip that the difference is noise.
Do blowouts go under?
Everyone says yes: the clock runs, the starters sit, the scoring stops. The archive says the opposite.
Bar = 95% confidence range, tick = measured rate. The faint line is 50% (pure chance); the green line is 52.38%, the break-even rate at −110 once the hold is included. Bars falling short of it describe historical rates below break-even.
Games with a spread of 28 or more went over 54.5% of the time, and the effect strengthens the more lopsided the game gets — 50.6% at 28–32, rising to 59.9% at 42+. That dose-response is what separates a real effect from noise, and it is the test wind and altitude both failed.
So why isn't this a play? Because it lives almost entirely in games against FCS opponents (58.1% there against 51.3% versus FBS teams), and because it does not survive dropping a single season: without 2018 the confidence range falls to 49.7%, and without 2025 to 52.1% — both under the 52.38% needed. Two of nine years carry it. We publish the direction because it is real and it contradicts what everyone repeats; we don't publish it as a bet because it isn't one.
Situations people bet on
Home field, neutral sites, conference play, September cupcakes. All measured from the home side against the spread.
Nothing here clears the bar. The home side has covered 49.2% of true home games across 7,706 of them — which is the market doing its job, not home field being worthless.
Can a public rating beat the closing line?
Two of them, tested independently, both land on the number.
Bar = 95% confidence range, tick = measured rate. The faint line is 50% (pure chance); the green line is 52.38%, the break-even rate at −110 once the hold is included. Bars falling short of it describe historical rates below break-even.
Pregame Elo disagreeing with the spread produces 48–50% at every threshold. That matches what SP+ did (49.8% over 1,589 games). Two independent public ratings both landing on the number is better evidence than either alone: the college market has already priced what free ratings know.
Does the line movement tell you anything?
Following the steam — backing whichever side the number moved toward between the open and the close.
Bar = 95% confidence range, tick = measured rate. The faint line is 50% (pure chance); the green line is 52.38%, the break-even rate at −110 once the hold is included. Bars falling short of it describe historical rates below break-even.
Coin flips at every size of move, on 2,150 games. Fading the steam is the mirror image and equally unprofitable. Movement is still worth watching for what it crossed — a line going through 3 or 7 changes far more outcomes than one moving through 9, which is what the key-number weighting measures.
What to take from this
The college spread market is priced well. That is not a disappointing answer — it is the one that tells you where the remaining value is: in the price you pay rather than the side you pick. Shopping a half-point around a key number moves more games than any of the angles on this page, and the hold on a futures market is a cost you can see and avoid before anyone is right about anything.