There is a band of outcomes where a moneyline ticket pays and a touchdown spread does not. It is the win by one through six. Every argument for paying up on a favourite instead of laying the points is an argument about how often a game finishes inside that band, and that is a measurable thing rather than a matter of taste.
Across the 888 FBS games of the 2025 regular season, 23.99 percent finished with a margin of six or fewer. Roughly one game in four. That number is the reason moneylines on modest favourites are priced the way they are, and it is also the number most people carry into an opening Thursday without checking whether openers behave like the rest of the calendar.
They do not.
Bobby Dodd Stadium, Atlanta. Opening night games get bet as coin flips waiting to happen. The 2025 season says the opposite. Photo: JJonahJackalope, CC BY-SA 4.0, via Wikimedia Commons.
Every FBS regular season game from the 2025 ESPN scoreboard with a final score, 888 of them, bucketed by absolute margin. College football has no ties, so every game lands somewhere.
| Margin of victory | Games | Share |
|---|---|---|
| 1 to 6 | 213 | 23.99% |
| 7 to 13 | 170 | 19.14% |
| 14 to 20 | 156 | 17.57% |
| 21 or more | 349 | 39.30% |
| All games | 888 | 100% |
Two things fall out of that table immediately. The first is that a quarter of the sport is a one score game. The second is that 39.30 percent of it is a three score game, which is a much larger share than basketball or baseball bettors expect when they arrive here from another sport. College football is not a close sport. It is a bimodal one, and the middle of the distribution is thin.
Now take the same table and restrict it to the 96 games played in week one.
| Margin of victory | All 2025 games | Week one only |
|---|---|---|
| 1 to 6 | 23.99% | 13.54% |
| 7 to 13 | 19.14% | 13.54% |
| 14 to 20 | 17.57% | 22.92% |
| 21 or more | 39.30% | 50.00% |
| Games | 888 | 96 |
Half of the openers were decided by three scores. The one score share was cut almost in half, from 23.99 percent to 13.54 percent. The received wisdom about opening week, that nobody knows anything yet and everything is a coin flip, is the opposite of what the scoreboard produced. Openers are scheduled to be mismatches and they play out as mismatches.
That has a direct consequence for the moneyline versus spread question. The band where the moneyline earns its extra juice, the win by one to six, is the least likely band in week one. If the favourite is going to win, week one says it is more likely to win by 21 than by 5.
The other half of the question is opponent quality. Take the 873 non neutral site games and keep the ones where the home team finished the season with a winning record and the visitor finished under .500. That is 236 games.
| Outcome | Games | Share |
|---|---|---|
| Home won by 21 or more | 130 | 55.08% |
| Home won by 14 to 20 | 48 | 20.34% |
| Home won by 7 to 13 | 33 | 13.98% |
| Decided by 1 to 6 | 25 | 10.59% |
| Visitor won outright | 5 | 2.12% |
The winning host won 231 of 236. Tighten it to a host with eight or more wins against a visitor with four or fewer and the sample is 158 games, the host won 157 of them, and only 8.86 percent of the set finished inside six points. Tighten it again to nine or more against three or fewer and the host went 66-0 with an average margin of 34.85.
Those last two rows are where the honesty has to start, because they are conditioned on information that did not exist when the games were played. A team is 9-3 partly because it won these games. That is circular and it inflates the rate. The 236 game version is less circular than the 66 game version and all three are more circular than a bettor would like. The number that survives that objection best is the margin distribution, not the win rate.
A moneyline on a favourite is a purchase of exactly one thing: the outcomes where the team wins but does not cover. Everything else is identical to the spread ticket. So the value of that purchase is the probability of the buy zone, and this data prices it.
| Situation | Games | Favourite wins inside six |
|---|---|---|
| All FBS games | 888 | 23.99% of all games |
| Week one | 96 | 13.54% of all games |
| Winning host, losing visitor | 236 | 9.75% won by 1 to 6 |
| Host 8W or better, visitor 4W or worse | 158 | 8.86% inside six either way |
Read the last two rows carefully. In the situation that looks most like a season opening home favourite against a team coming off a bad year, fewer than one game in ten lands in the band the moneyline is bought for. The moneyline in that spot is not paying for a likely outcome. It is paying to remove a tail.
Whether that is worth the price is a question about the price, and there are no prices in this study. What the study can say is the shape: the buy zone is smaller in week one than in any other week, smaller again when the visitor is bad, and the thing that actually happens in these games is a blowout, 55.08 percent of the time by three scores or more.
There is a mirror version of this for the underdog side. A visitor with a losing record won 5 of 236 at a winning host, 2.12 percent. Against that, the same visitor kept the game inside six 10.59 percent of the time. A dog bettor in that spot is buying five times more cover equity than win equity, and the moneyline dog is the worse of the two tickets by a wide margin.
One season is one season. 888 games is a reasonable sample for a margin distribution and a thin one for a 66 game subset. Week one in 2025 was scheduled the way week one is usually scheduled, with a handful of marquee neutral site games sitting on top of a lot of arranged mismatches, and if a season ever arrives where openers are scheduled competitively the 13.54 percent will move. The direction of the finding is more durable than the digits.
How This Was Built
- Game sample
- Every 2025 FBS regular season game returned by the ESPN college football scoreboard endpoint for season type 2, weeks 1 through 16, group 80, with a final status. 888 games. Postseason and FCS only games are excluded.
- Margin
- The absolute difference of the two final scores, overtime included at face value. College football has no ties so every game falls in exactly one bucket.
- Week one
- The 96 games ESPN assigns to week 1 of the 2025 regular season, including the games played before Labor Day weekend.
- Non neutral sample
- 873 of the 888 games, dropping every event ESPN flags as a neutral site, so that home and visitor mean something.
- Team records
- Computed from this same 888 game set, so they are FBS regular season records rather than the full published record. A team's record is applied to every one of its games including the ones that produced it.
- Buckets
- Winning host means more wins than losses in the sample. Losing visitor means more losses than wins. The 8W and 9W cuts are raw win counts in the sample.
What is not checked, and cannot be:
- No prices anywhere. A 9.75 percent buy zone is worth a different amount at -140 than at -260 and none of that is measured here.
- End of season records are applied backwards to every game. A host is 9-3 partly because it won the games in the sample, which inflates the win rate in the tightest buckets. The 66-0 line is the worst offender and should be read as an illustration, not an estimate.
- One season only. The week one figure rests on 96 games and a binomial standard error at 13.54 percent on 96 observations is 3.5 points.
- Schedule construction is not controlled. Week one mismatches are chosen, not drawn, and a differently scheduled opening week would move the number.
- Nothing is adjusted for injuries, transfers, coaching turnover or the roster churn that makes last season's record a rough proxy for this season's team.
- Neutral site games are dropped from the host buckets but kept in the overall margin table, so the two are not on identical samples.
- Nothing here says which side of any particular game to take.