What Are the Key Numbers in NFL Betting?
Three and seven, then 6, 10, 14 and 4. Across 7,276 NFL games since 1999, 15.1% ended by exactly 3 and 9.0% by exactly 7.
September 27, 2026 at 5:21 PM EDT
6 min read
The key numbers in NFL betting are the final margins that come up far more often than their neighbours: 3 above all, then 7, and after those 6, 10, 14 and 4. They exist because football is scored in 3s and 7s. A field goal decides a lot of close games, a converted touchdown decides a lot of others, and combinations of the two (10, 14, 17) stack up behind them.
We counted every completed NFL game in the nflverse results file from 1999 through the 2025 season, 7,276 of them including playoffs. 1,100 ended with a margin of exactly 3 points, 15.1% of all games. 657 ended by exactly 7, another 9.0%. One game in four finished on one of those two numbers. No other margin tops 6%.
How often does each margin actually happen?
Here is the full top of the distribution, winner's margin regardless of who was favoured:
| Final margin | Games | Share |
|---|---|---|
| 3 | 1,100 | 15.1% |
| 7 | 657 | 9.0% |
| 6 | 436 | 6.0% |
| 10 | 401 | 5.5% |
| 4 | 355 | 4.9% |
| 14 | 355 | 4.9% |
| 1 | 303 | 4.2% |
| 2 | 295 | 4.1% |
| 8 | 271 | 3.7% |
The drop from 3 to everything else is the thing to absorb. A margin of 3 is roughly three times as common as a margin of 4, 1 or 2. That is not a quirk of a few seasons; 3 stayed far ahead of every other margin in both eras we split the file into (see the 2015 table below).
A margin only matters to a spread bettor when it lines up with the spread. The more useful count is conditional: in the games where the market actually set the favourite at a given number, how often did the favourite win by exactly that number? In the 1,153 games listed at exactly 3, the favourite won by exactly 3 in 9.0% of them. In the 491 games listed at exactly 7, the favourite won by exactly 7 in 6.5%. Those are the numbers a half point is fighting over.
What is a half point worth around 3 and 7?
A half point across a key number changes the result of every game that lands on it. Take an underdog at +2.5 versus +3. If the favourite wins by exactly 3, the +2.5 ticket loses and the +3 ticket is refunded (both books' house rules settle a whole-number spread that lands exactly as a push). Move to +3.5 and that same game becomes a win.
Put a price on it with one simplifying assumption: the line is fair at 3, so apart from the games that land on 3 the underdog wins and loses equally often. With 9.0% of games landing on 3, that leaves 45.5% wins and 45.5% losses for the +3 ticket.
- +2.5 at -110: win 45.5%, lose 54.5%. Per $1 risked: 0.455 × 0.909 − 0.545 = −$0.132.
- +3 at price −X: win 45.5%, push 9.0%, lose 45.5%. Per $1 risked: 0.455 × (100/X) − 0.455.
Set the two equal and solve: 100/X = 1 − 0.132/0.455 = 0.711, so X ≈ 141. Under this model you could pay up to about −140 for +3 and be no worse off than taking +2.5 at −110. Run the same arithmetic at 7 with its 6.5% landing rate and the ceiling is about −130.
Compare a half point that crosses nothing important. In the same games listed at 3, the favourite won by exactly 4 only 2.9% of the time. The same model says moving onto 4 is worth paying about −118 at most. That gap is why a flat price for every half point cannot be right. A book that charges the same to move you from 3.5 to 3 as from 4.5 to 4 is overcharging on the dead number or undercharging on the key one, and the numbers say which.
Two warnings on that model. It assumes the line was right; if you are buying a half point on a bet with no edge, you are paying less to lose less. And the landing rates are historical averages. A low-total game between two field-goal-heavy offences lands on 3 more often than a shootout does, which is why the rate should be estimated for the game in front of you, not borrowed from the league.
Did the 2015 extra-point change weaken 7?
In May 2015 NFL owners voted 30-2 to move the extra-point kick back to the 15-yard line for the 2015 season, leaving two-point tries at the 2 and letting the defence score two points on a return. A longer kick means more misses, and more coaches going for two, and both of those break up 7-point margins into 6s and 8s.
The results file shows exactly that shift, with the usual caution that correlation over two eras is not proof of cause:
| Margin | 1999–2014 (4,248 games) | 2015–2025 (3,028 games) |
|---|---|---|
| 3 | 15.3% | 14.8% |
| 6 | 5.3% | 6.9% |
| 7 | 9.3% | 8.7% |
| 8 | 3.3% | 4.3% |
| 10 | 6.1% | 4.8% |
| 14 | 4.7% | 5.2% |
Seven is still the second key number. It is a little less dominant, and 6 and 8 took most of what it lost. We did not measure kick make rates or two-point attempt rates in this piece; those would be the direct test of the mechanism. Three barely moved, which fits: a field goal is a field goal whatever happens after a touchdown.
Where the key numbers show up in your betting
Line shopping. The difference between −3 at −120 and −3.5 at −105 at two different books is not five cents of vig. It is the 9% of games that land on 3. Our spread betting guide explains how to read both halves of that price; the point here is that the half point and the juice have to be compared together.
Line moves. When a spread moves from 2.5 to 3, it has crossed the most valuable spot on the board. A move from 3 to 3.5 is a much bigger event than 4 to 4.5. That is also why a book reluctant to leave 3 will usually move the price attached to it (−105, −115, −125) before it moves the number.
Middles. If you took +3.5 early and the line later moves to −2.5 the other way, a final margin of 3 wins both bets. The frequency table above is the probability that makes that position worth holding; the Academy's guide to middles covers the mechanics.
Teasers. Teasing through 3 and 7 is the whole logic of the most famous NFL teaser; see NFL teasers explained, next in this series.
What the table does not tell you
It does not tell you who will cover. The share of games landing on 3 says nothing about which side of 3 a particular game falls. It also includes overtime results, as the books' settlement rules do; a game decided by an overtime field goal is a 3-point margin for settlement.
Totals cluster too, but far more weakly, because a total is the sum of two teams' scoring rather than the difference. In the same file the most common combined score, 37, came up in 3.8% of games, with 41, 51 and 44 close behind at 3.7% to 3.8%. No total comes near the 15.1% that 3 claims among margins, so a half point on a total is worth much less than a half point on 3.
Sports betting is legal only in some states and only for adults 21 and over in most of them. The frequencies here describe history; they are not predictions for any game.