NFL market prices — what each bet is actually worth
A book posts forty markets on a game and publishes the odds for none of them. These are the fair prices, measured over 2,827 games, so you can check what you are being offered. Every figure carries its sample and a confidence interval.
Winning margin
Margins of 3 and 7 are the two spikes that make this market worth knowing. They move with the spread: a margin of 3 is near 17% in a close game and near 10% when the favourite is giving 11 or more.
By exact margin — all games
| Margin | Happens | Fair price | Want at least | Sample |
|---|---|---|---|---|
| Exactly 3key number | 14.8% 13.5%–16.1% | 6.78 | 7.41 | 2,827 |
| Exactly 7key number | 8.5% 7.5%–9.5% | 11.83 | 13.36 | 2,827 |
| Exactly 6 | 6.7% 5.9%–7.7% | 14.88 | 17.08 | 2,827 |
| Exactly 14 | 5.1% 4.3%–5.9% | 19.77 | 23.2 | 2,827 |
| Exactly 2 | 4.7% 4%–5.5% | 21.42 | 25.31 | 2,827 |
| Exactly 4 | 4.7% 4%–5.6% | 21.1 | 24.9 | 2,827 |
| Exactly 10 | 4.7% 4%–5.5% | 21.42 | 25.31 | 2,827 |
| Exactly 1 | 4.6% 3.9%–5.4% | 21.91 | 25.95 | 2,827 |
| Exactly 5 | 4.5% 3.8%–5.3% | 22.26 | 26.39 | 2,827 |
| Exactly 8 | 4.4% 3.7%–5.2% | 22.62 | 26.86 | 2,827 |
| Exactly 17 | 3.5% 2.9%–4.3% | 28.27 | 34.28 | 2,827 |
| Exactly 16 | 2.3% 1.8%–3% | 42.83 | 54.37 | 2,827 |
| Exactly 18 | 2.3% 1.8%–3% | 42.83 | 54.37 | 2,827 |
| Exactly 20 | 2.2% 1.7%–2.8% | 46.34 | 59.4 | 2,827 |
By band
| Margin band | Happens | Fair price | Want at least | Sample |
|---|---|---|---|---|
| 1 to 6 | 39.9% 38.1%–41.8% | 2.5 | 2.62 | 2,827 |
| 7 to 12 | 23.3% 21.8%–24.9% | 4.3 | 4.6 | 2,827 |
| 13 to 18 | 17% 15.6%–18.4% | 5.89 | 6.39 | 2,827 |
| 19 to 24 | 10% 9%–11.2% | 9.99 | 11.16 | 2,827 |
| 25 or more | 9.4% 8.4%–10.6% | 10.59 | 11.88 | 2,827 |
How the margin moves with the spread
The same market, conditioned on how big the handicap is. Quoting one number for every game would be wrong at both ends of the board.
| Spread | Games | Margin 3 | Margin 7 | 1 to 6 | 25+ |
|---|---|---|---|---|---|
| 373 | 16.9% | 9.1% | 43.7% | 6.2% | |
| 1,096 | 17% | 9.8% | 44.2% | 7.3% | |
| 747 | 12.7% | 8.2% | 39% | 8.8% | |
| 363 | 12.9% | 7.7% | 35.8% | 10.5% | |
| 248 | 10.5% | 3.6% | 24.6% | 24.2% |
Odd or even — the market that looks like a coin flip
A scoreless quarter counts as EVEN, and 9% of first quarters are scoreless. Against that, touchdowns and field goals are both odd numbers, which pushes full-game totals the other way. The two effects do not cancel.
| Market | Odd | Fair | Even | Fair | Scoreless |
|---|---|---|---|---|---|
| Game total | 53% | 1.89 | 47% | 2.13 | 0% |
| 1st half | 49.9% | 2.01 | 50.1% | 1.99 | 0.2% |
| 2nd half | 50.1% | 1.99 | 49.9% | 2.01 | 0.3% |
| 1st quarter | 50.9% | 1.97 | 49.1% | 2.04 | 9.4% |
| 2nd quarter | 49.6% | 2.02 | 50.4% | 1.98 | 1.7% |
| 3rd quarter | 51.3% | 1.95 | 48.7% | 2.05 | 8.5% |
| 4th quarter | 48.7% | 2.05 | 51.3% | 1.95 | 4.7% |
Worked example, from a real board. Sportsbet priced the game total at 1.85 odd / 1.88 even. Against 53% / 47% that makes odd a -1.1% bet and even a -6.2% bet — five points apart, on two sides of the same market, priced as though they were the same thing. Neither is a winning bet. One is much worse.
Quarter shape
Highest-scoring quarter excludes ties, which are common enough to matter — 10.6% of games have two quarters level at the top, and most books settle those by dead heat rules rather than paying in full.
Highest-scoring quarter
| Quarter | Happens | Fair price | Want at least | Sample |
|---|---|---|---|---|
| Q1 | 9.8% 8.7%–10.9% | 10.23 | 11.46 | 2,772 |
| Q2 | 35.9% 34.1%–37.7% | 2.79 | 2.93 | 2,772 |
| Q3 | 12.2% 11%–13.4% | 8.23 | 9.1 | 2,772 |
| Q4 | 31.6% 29.9%–33.4% | 3.16 | 3.34 | 2,772 |
Scoreless quarters, and both teams scoring
| Outcome | Happens | Fair price | Want at least | Sample |
|---|---|---|---|---|
| Q1 scoreless | 9.4% 8.3%–10.5% | 10.66 | 11.98 | 2,772 |
| Q2 scoreless | 1.7% 1.3%–2.3% | 57.75 | 76.42 | 2,772 |
| Q3 scoreless | 8.5% 7.5%–9.6% | 11.8 | 13.34 | 2,772 |
| Q4 scoreless | 4.7% 4%–5.6% | 21.16 | 25.02 | 2,772 |
| Any quarter scoreless | 22.4% 20.9%–24% | 4.47 | 4.8 | 2,772 |
| Both teams score in every quarter | 7.6% 6.7%–8.7% | 13.08 | 14.89 | 2,772 |
Order of scoring — first, last, and the race markets
Who scores first, who scores last, who gets to 5, 10, 15 or 20 points first. A book posts all of these and they are among the least examined numbers on the board. Priced here from 2,667 games, 2016–2026, every price conditioned on the posted handicap and — where it matters — the posted total.
The board, at seven handicaps
Probability, with the fair decimal price beneath it. Totals held at 44.5. The last column is how much the handicap improves on simply knowing how often the thing happens — measured on seasons the model had never seen.
| Market | Home −10 | Home −7 | Home −3 | Pick’em | Home +3 | Home +7 | Home +10 | Skill |
|---|---|---|---|---|---|---|---|---|
| Home team to score first | 61.5%1.63 | 58.1%1.72 | 53.4%1.87 | 49.8%2.01 | 46.2%2.16 | 41.5%2.41 | 38.1%2.63 | 1.8% |
| Home team to score last | 58.8%1.7 | 57.1%1.75 | 54.7%1.83 | 53%1.89 | 51.2%1.95 | 48.8%2.05 | 47%2.13 | 0.5% |
| Home team to score first in the 2nd half | 61.7%1.62 | 57.1%1.75 | 50.9%1.96 | 46.2%2.16 | 41.6%2.41 | 35.6%2.81 | 31.4%3.18 | 3.1% |
| First scoring play is a touchdown | 61%1.64 | 61%1.64 | 61.1%1.64 | 61.2%1.64 | 61.2%1.63 | 61.3%1.63 | 61.3%1.63 | 0.2% |
| Team that scores first wins the game | 63%1.59 | 62.5%1.6 | 61.9%1.61 | 61.5%1.63 | 61%1.64 | 60.4%1.66 | 59.9%1.67 | 0.1% |
| Both teams score | 98.2%1.018 | 98.2%1.018 | 98.2%1.018 | 98.2%1.018 | 98.2%1.018 | 98.2%1.018 | 98.2%1.018 | -1% |
| Home team races to 5 | 66.3%1.51 | 61.8%1.62 | 55.4%1.81 | 50.4%1.98 | 45.5%2.2 | 39.1%2.56 | 34.5%2.9 | 3.6% |
| Away team races to 5 | 33.6%2.98 | 38.1%2.62 | 44.5%2.25 | 49.4%2.02 | 54.3%1.84 | 60.8%1.64 | 65.4%1.53 | — |
| Neither side reaches 5 | 0.1%1712.65 | 0.1%1247.92 | 0.1%818.34 | 0.2%596.42 | 0.1%818.34 | 0.1%1247.92 | 0.1%1712.65 | -0.6% |
| Home team races to 10 | 69%1.45 | 63.4%1.58 | 55.4%1.8 | 49.2%2.03 | 43%2.32 | 35.2%2.84 | 29.7%3.37 | 5.4% |
| Away team races to 10 | 30.6%3.27 | 36.1%2.77 | 44.1%2.27 | 50.3%1.99 | 56.4%1.77 | 64.4%1.55 | 69.8%1.43 | — |
| Neither side reaches 10 | 0.5%221.07 | 0.5%210.34 | 0.5%196.85 | 0.5%187.3 | 0.5%196.85 | 0.5%210.34 | 0.5%221.07 | -0.1% |
| Home team races to 15 | 75.8%1.32 | 68.5%1.46 | 57%1.75 | 47.7%2.1 | 39.2%2.55 | 28.7%3.49 | 21.9%4.57 | 10.7% |
| Away team races to 15 | 22.6%4.42 | 29.6%3.38 | 40.3%2.48 | 48.8%2.05 | 58%1.72 | 69.4%1.44 | 76.6%1.31 | — |
| Neither side reaches 15 | 1.5%65.09 | 2%50.78 | 2.7%36.54 | 3.5%28.6 | 2.7%36.54 | 2%50.78 | 1.5%65.09 | 3.5% |
| Home team races to 20 | 74.2%1.35 | 65.9%1.52 | 52.8%1.89 | 42.3%2.36 | 35.3%2.83 | 25.7%3.89 | 19.3%5.17 | 12.6% |
| Away team races to 20 | 19.2%5.21 | 25.5%3.92 | 35.1%2.85 | 42.1%2.37 | 52.6%1.9 | 65.7%1.52 | 74%1.35 | — |
| Neither side reaches 20 | 6.6%15.1 | 8.6%11.59 | 12.1%8.24 | 15.5%6.44 | 12.1%8.24 | 8.6%11.59 | 6.6%15.1 | 6.1% |
What the backtest actually says
Each market trained on every prior season and graded on the next, never on itself. Brier score against a constant predictor fed the training-set base rate — which is the matched baseline here, not 50%, and not the market.
| Market | Graded | Landed | Brier | Base rate only | Skill | Calibration |
|---|---|---|---|---|---|---|
| Home team to score first | 1,662 | 52.4% | 0.2453 | 0.2499 | 1.8% | 1.06 |
| Home team to score last | 1,662 | 52.8% | 0.2490 | 0.2503 | 0.5% | 0.63 |
| Home team to score first in the 2nd half | 1,658 | 48.9% | 0.2423 | 0.2501 | 3.1% | 1.11 |
| First scoring play is a touchdown | 1,660 | 61.6% | 0.2362 | 0.2367 | 0.2% | 0.71 |
| Team that scores first wins the game | 1,656 | 62.1% | 0.2356 | 0.2358 | 0.1% | 0.39 |
| Both teams score | 1,662 | 98.3% | 0.0173 | 0.0172 | -1% | 0.61 |
| Home races to 5 | 1,661 | 53.6% | 0.2407 | 0.2497 | 3.6% | 1.08 |
| Neither reaches 5 | 1,662 | 0.1% | 0.0006 | 0.0006 | -0.6% | 1.21 |
| Home races to 10 | 1,654 | 53.4% | 0.2360 | 0.2494 | 5.4% | 1.06 |
| Neither reaches 10 | 1,662 | 0.5% | 0.0048 | 0.0048 | -0.1% | 0.98 |
| Home races to 15 | 1,609 | 53.5% | 0.2225 | 0.2490 | 10.7% | 1.06 |
| Neither reaches 15 | 1,662 | 3.2% | 0.0299 | 0.0309 | 3.5% | 1.29 |
| Home races to 20 | 1,463 | 54.4% | 0.2170 | 0.2484 | 12.6% | 1.07 |
| Neither reaches 20 | 1,662 | 12% | 0.0992 | 0.1056 | 6.1% | 1.18 |
14 tests were run, so about 0.7 of them would look good by chance alone at the 5% level. Every one is listed above, including the ones that found nothing. Every market above beat a flat base rate out of sample, though several by very little. Overconfidence is corrected by Platt scaling fitted on the walk-forward predictions rather than by discarding the market, which is why some calibration figures sit above 1.00. The curve fits worst on Neither reaches 20 (chi-squared 30.5 across 7 handicap bands, where about 7 is what a correct curve gives) — treat that row with more suspicion than the others.
Forty-six international games are excluded: the two data sources disagree about which side is at home, and guessing would flip the handicap against every outcome in them.
What this page still cannot price, and why
These need something the data does not carry: which player scored. Everything else on a book's board is priced above.
| Market | What it would need |
|---|---|
| First and last scorer | scoring plays with player attribution |
| Anytime touchdown doubles | player-level joint probabilities |
How to use a fair price
- Convert the offer. A decimal price of 6.50 implies 100 ÷ 6.50 = 15.4%.
- Compare with the column above. Margin 3 is 14.8%, so 6.50 is roughly fair and 7.50 is value.
- Use the lower bound on thin markets. The "want at least" column is the fair price at the pessimistic end of the interval. On a market measured off a few hundred games, that is the number worth holding out for.
- Remember the dead heat. 10.6% of games have two quarters tied as highest-scoring, and most books halve the stake rather than paying in full. The fair prices above are for an outright win, so shade them.
More NFL
Built 2026-10-18 from 2,827 games. Fair prices are what the outcome historically costs, not a prediction for any one game, and not a bookmaker quote. Break-even at standard −110 pricing is 52.4%.