Learn
The One Season Trap: how many bets to prove an edge in the NFL and beyond
If you want to know how many bets to prove an edge, the One Season Trap table gives the answer for common win rates at -110, and it is far more than one NFL season. Know the break-even bar, count the bets each edge needs, and grade every bet against the closing line in the meantime.
One NFL season feels like a lot of football. For judging a betting model it is a small sample, and most of the trap is that it doesn't feel small. The One Season Trap is three moves from the video, and the table below is the part you will want to keep.
- Know the bar. At -110 you need to win about 52.4% of the time to break even (moneyline odds).
- Count the bets. Work out how many bets your claimed edge needs before luck stops being the likely explanation.
- Check the close. Until you have that many, judge each bet by whether its price beat the closing line.
Step 1. Know the bar
At -110 you risk 110 to win 100, so the break-even win rate is 110 divided by 210, or 52.38% (moneyline odds). Every claimed edge has to clear that bar first.
It is a higher bar than it looks. In a study of 2,560 NFL games from 2002 to 2011, Szalkowski and Nelson found that home teams beat the spread only about 47% of the time, and that a simple home-underdog rule would have won about 53.5%, just above the 52.38% break-even. Real edges against the spread tend to be a point or two above the bar, not ten.
Step 2. Count the bets
A win rate over a small number of bets swings a lot. The table shows, for a bettor whose true skill is the win rate on the left, how often the record would still sit below break-even after a given number of bets, and how many bets it takes before the record sits about two standard errors above the bar. All numbers are hypothetical, from the normal approximation to the binomial.
| True win rate at -110 | Below break-even after 60 bets | After 100 | After 272 (one NFL season, every game) | After 1,000 | Bets for a 2-standard-error margin | NFL seasons of every game |
|---|---|---|---|---|---|---|
| 53% | 46% | 45% | 42% | 35% | about 26,000 | about 96 |
| 54% | 40% | 37% | 30% | 15% | about 3,800 | about 14 |
| 55% | 34% | 30% | 19% | 5% | about 1,440 | about 5.3 |
| 56% | 29% | 23% | 12% | 1% | about 750 | about 2.8 |
| 57% | 24% | 18% | 6% | under 1% | about 460 | about 1.7 |
The season length is 17 games for each of 32 teams, which is 272 games. The bets-needed column uses the standard error of a win rate, the square root of p times one minus p over n, and asks for the gap to break-even to be twice that size.
Read the 55% row slowly. A bettor that good, betting every game of a season, still finishes the year below break-even about one time in five. Over a hot month of about 60 bets, it is about one time in three. You can rerun any row with your own numbers using the binomial distribution, and the logic doesn't change: small samples hide real edges and invent fake ones.
Step 3. Check the close while you wait
You can't wait 1,400 bets to find out whether a model is any good. The closing line is the market's final price, and betting markets are hard to beat: Wunderlich and Memmert note that betting odds are frequently found to outperform mathematical models. So if the prices you take keep beating the close, that is a meaningful signal that shows up on every single bet. Hubáček and Šír go further and show a model can profit while being worse at prediction than the market, as long as its errors are not correlated with the market's.
The routine:
- Log every bet with the price you took and the closing price for the same side.
- Mark each one beat, matched or missed. Our closing line value calculator does the math.
- After every 50 bets, look at the share you beat and the average size of the beat, not at the record.
A season log you can copy
| Week | Bets | Record | Win rate | Bets that beat the close | Share beating the close |
|---|---|---|---|---|---|
| 1 | |||||
| 2 | |||||
| Season total |
Add one rule at the top before week 1: which bets the model takes, at which price, and how much you stake. Don't change it mid-season.
What to watch out for
- Judging by the record. A 9-3 month and a 4-8 month can both come from the same 55% model, as the break-even math and the table above imply.
- Changing the rule after a losing run. That resets your sample to zero.
- Testing many models and keeping the best season. One of several no-edge models will often have a good year by luck.
- Stake creep. A hot start talks you into bigger bets on a skill you haven't proven. Keep the unit fixed for the whole test.
How to tell it worked
By the end of a season you won't have proof from the record, and that's expected. What you should have is a clear closing-line number across a few hundred bets and a record that sits inside the range the table predicts for your claimed skill. If your prices rarely beat the close, the record was probably luck in either direction. Every Quant+ model publishes a scorecard with how often it beats the close, walk-forward by season, so you can apply the same test to ours.
21+. For informational purposes only. Nothing is guaranteed. If gambling stops being fun, call 1-800-GAMBLER.
Questions people ask
What win rate do I need to break even at -110?
About 52.38%, because you risk 110 to win 100. Anything below that loses money over time, before any edge comes into it.
Is a winning NFL season proof that a model works?
No. In a hypothetical where a model truly wins 55% against the spread and bets all 272 regular-season games, it still finishes below break-even about one time in five.
How many bets does a 55% bettor need?
About 1,400 bets for the result to sit two standard errors above break-even, which is more than five NFL seasons of betting every game. Smaller edges need far more.
What can I check before I have that many bets?
Closing line value. If the prices you take keep beating the closing line, that is evidence of an edge you can measure on every bet, long before the win-loss record settles.
Sources
- The Performance of Betting Lines for Predicting the Outcome of NFL Games, arXiv (Szalkowski, Nelson)
- Moneyline odds, Wikipedia
- The Betting Odds Rating System, PLOS ONE (Wunderlich, Memmert)
- Binomial distribution, Wikipedia
- Beating the market with a bad predictive model, arXiv (Hubáček, Šír)
