Key numbers
In basketball a 4-point win and a 5-point win are about equally likely. NFL scoring comes in field goals and touchdowns, so some final margins turn up far more often than their neighbors. These are key numbers, and they decide what half a point is worth.
How often games land on 3 and 7
In the NFL margin table we use, built from regular-season games since 2015 and smoothed:
- About 14.6% of games end with one team winning by exactly 3, about 1 in 7.
- About 8.7% end by exactly 7, about 1 in 11.
- Next come 6 and 10, and 10 lands almost three times as often as 9.
3 and 7 lead every other margin, which is why NFL bettors talk about them constantly.
What a half-point is worth
A half-point, the hook, is worth the chance that the game lands exactly on the number it crosses. Moving from +2.5 to +3 turns a 3-point loss into a push; moving from +3 to +3.5 turns that push into a win. Across 3, that is a lot of games. Across 8 or 9, it is a sliver. A spread tool that treats every half-point the same is wrong in exactly the places that matter most.
Books know this too. Around 3 they often move the price before they move the number: Bears −3 at −120 is a different bet from Bears −3 at −110, and the market may sit there for days rather than cross to 3.5. When the number finally does cross 3, it is a big move, not a small one.
Buying points
Many books let you buy a half-point for a worse price. Whether that is worth it depends on only two things: how often the game lands on the number you cross, and what the extra price costs you. Buying onto or off 3 can be worth a lot; buying through 5 or 9 almost never is. Books that sell points usually charge more around 3 and 7 for the same reason.
The same logic applies when you shop. Bears −2.5 at one book and −3 at another are not the same bet at slightly different prices; the half-point can be worth more than the price gap. It is also why a price graded at one number doesn't carry over to another: a bet off the number is a different bet.
What it means for pricing
A model that prices spreads off a smooth bell curve, a normal distribution, smooths the chances across margins, so it undervalues the half-points across 3 and 7 and overvalues the rest. Pricing from measured margins, like the table above, gets them right. When you compare alternate spreads, or any tool's fair price for one, check that the half-points across 3 and 7 are priced well above the others. Whatever a game's verdict, the key numbers matter on every spread you bet.
Example. You like the Lions, a 3-point underdog. Your book offers Lions +3 at −110, or Lions +3.5 if you buy the half-point at −125 (illustrative prices, not live data). In the same table, among games with a spread near 3, the favorite won by more than 3 in 44.83%, by exactly 3 in 9.30%, and by less than 3 (or lost) in 45.87%. That 9.30% is lower than the 14.6% of all games above because only the favorite winning by exactly 3 pushes a Lions +3, while the 14.6% counts either team winning by 3.
| Lions +3 at −110 | Lions +3.5 at −125 | |
|---|---|---|
| Wins | 45.87% | 45.87% + 9.30% = 55.17% |
| Pushes | 9.30% | none |
| Loses | 44.83% | 44.83% |
| Profit on a $100 win | $90.91 | $80.00 |
| Average result per $100 | 0.4587 × $90.91 − 0.4483 × $100 = −$3.13 | 0.5517 × $80 − 0.4483 × $100 = −$0.69 |
Both are slightly negative here because you pay the vig on what is close to a coin flip, and a table average is not a read on this game. The point is the gap: the hook across 3 is worth −$0.69 − (−$3.13) = about $2.44 per $100, even at −125. The same purchase from +4 to +4.5 buys only the games that land exactly on 4, about 2.7% of games lined near 4 in the same data, under a third as many.