Key Numbers and Parlays: The Three-Point Problem

NFL scoring is not spread evenly across the scoreboard. In 3,028 games from 2015 through the 2025 playoffs, 449 finished with a three-point margin and 263 finished with a seven-point margin, which is 23.5 percent of every game on one of two numbers. The Best Bet on Sports has bet these markets live for more than twenty years with a verified $367,520+ in profit across all sportsbooks, and the single most expensive habit we see in parlay construction is treating every half point on a ticket as if it costs the same. It does not. On a parlay, the number you accept on each leg compounds.

A straight bet has one exposure to the key number. A three-leg parlay has three, and the ticket dies if any one of them lands wrong. That is the whole argument, but the size of the effect surprises people, so here is the arithmetic with the games attached.

Where Do NFL Games Actually Finish?

The margin distribution is the foundation of everything else on this page. Using the nflverse game file for 2015 through the 2025 playoffs, 3,028 games in total, here is how often each final margin occurred.

Three is not merely the most common margin. It occurs more often than the next two combined, and it is stable year to year rather than a quirk of one season. The per-season share of three-point games across this sample ran from 10.49 percent in 2019 to 16.90 percent in 2022, and the most recent season in the set, 2025, came in at 16.14 percent. There is no era in the sample where three stops mattering.

It is also worth knowing who wins those games, because it kills a common assumption. Of the 449 three-point finishes, the favorite won 262 and the underdog won 187. The key number is not a favorite's number or a dog's number. It is simply where football lands.

Why a Key Number Hurts a Parlay More Than a Single Bet

A parlay is a conjunction. Every leg has to be right, so every leg's tail risk adds to the ticket rather than averaging out against the others.

Across the same 3,028 games, the final score landed exactly half a point away from the closing spread in 4.79 percent of games. That is the set of games where moving the number by a single half point would have flipped the result. On one bet, a 4.79 percent flip zone is a rounding concern. Stack legs and it stops rounding.

Read the four-leg row. More than a third of four-leg spread tickets contain at least one leg that finished within a single point of the number. That is not a claim those tickets lost. It is a claim that on better than one ticket in three, the entire outcome turned on a half point or a point somebody accepted without negotiating.

This is also why our position on leg count has never moved. We wrote the longer version in how many legs a parlay should have, and the key-number data is an independent argument for the same conclusion, arrived at from a different direction.

How Often Does a Parlay Leg Land Exactly on the Number?

Pushes deserve their own section, because on a parlay a push does not lose. It removes the leg and recomputes the ticket at lower odds, which is better than a loss and worse than a win.

In our sample the closing spread was an exact whole number often enough to matter, and books hung the game on three in 433 of 3,028 games, or 14.30 percent. Seven came up 172 times, 5.68 percent. When the number is a whole number, the push becomes live.

A game hung on exactly three pushed 10.16 percent of the time. One leg in ten. Now compound it the way a parlay compounds everything else. A ticket built entirely out of three-point spreads carries a 19.29 percent chance of at least one push on two legs, 27.49 percent on three, and 34.86 percent on four. Across all spreads mixed together the per-leg push rate is 2.48 percent, so the same three-leg ticket built without regard to the number sits at 7.25 percent.

That gap, 27.49 percent against 7.25 percent, is entirely a consequence of which numbers you put on the ticket. It is not luck and it is not variance. It is selection. For what actually happens to the payout when a leg pushes, the mechanics are in what happens when a parlay leg pushes; this page is about how often you should expect to find out.

Is Half a Point Worth Buying on a Parlay Leg?

Sometimes, and the data says exactly when.

Compare the three closest numbers in the sample, looking at them from the favorite's side:

Moving a favorite from -3 to -2.5 buys nothing at all. You surrender the push and gain a fraction. Moving a favorite from -3.5 to -3 is the valuable direction, because it converts roughly a tenth of those games from a loss into a refunded leg. Moving an underdog from +2.5 to +3 does the same thing from the other side.

The general rule that falls out is short. Buy toward the key number, never away from it, and only when the half point actually crosses three or seven. Paying to move from -8 to -7.5 buys you nothing but vig, because nothing lives at 7.5. Paying to move from +2.5 to +3 buys a real slice of a real number.

On a straight bet that decision is worth a little. On a four-leg ticket where you are making it four times, it is worth four times as much, and it is being made against a payout structure that already carries more margin than a single bet does. If you want that margin math on its own, we laid it out in why most parlays lose.

Do Totals Have Key Numbers Too?

They do, but they are far weaker, and the practical implication is the opposite of what people assume.

In the same 3,028 games, the final combined score landed exactly on the closing total only 0.92 percent of the time. It landed within one point of the total 6.74 percent of the time and within three points 20.11 percent of the time. Compare that to the spread, where 7.27 percent of games finished within half a point of the number and 23.25 percent finished within three.

Total scoring is spread across far more possible outcomes than a margin is, so no single total carries anything like the weight that three carries on a spread. The consequence for ticket construction is that half points on totals are cheap in both directions, and paying up for them is usually wasted money. If you are going to spend on a half point somewhere on a parlay, spend it on the spread leg sitting next to three.

How to Build a Ticket That Respects the Numbers

Four working rules, all of them falling directly out of the tables above.

Count how many of your legs sit on or beside three. If more than one does, you are running a push-heavy ticket whether you meant to or not. That is not automatically wrong, but it should be a decision.

Take the whole number over the half point when you are on the right side of it. A refunded leg on a three-leg parlay is dramatically better than a lost one, and at 10.16 percent it is not a rare event.

Do not pay for half points that cross nothing. The only numbers worth buying through in professional football are three and, to a much smaller degree, seven. Everything else is the book selling you comfort.

Keep the leg count low enough that the flip zone stays small. At two legs, one in eleven tickets has a leg within a half point. At six, it is one in four. You can fix that with selection or you can fix it by building fewer legs, and building fewer legs costs nothing.

None of this makes a parlay a good bet by itself. It makes the difference between a parlay priced as carefully as you would price a single wager and one assembled out of whatever the app suggested. Our broader read on which structures survive the math is on the parlay betting strategy page, and our week-to-week board sits on NFL picks and football picks.

Get Tonight's Live Picks

Want tonight's live in-game picks delivered during the game, taken at numbers that move through three and seven while the market is repricing?

The Best Bet on Sports is the only live betting handicapping service limited on all six major U.S. sportsbooks (FanDuel, DraftKings, Caesars, BetMGM, Fanatics, ESPN BET) for winning too much during in-game action. Verified profit: $367,520+. Picks delivered via Email and Discord during games.

→ Get tonight's live picks: $199 first month — 1-Unit package, full live betting access → Try a live pick first — reserve your spot for tonight's pick

Compare every package on the buy sports picks page, see what a real sports handicapping service owes you for the money, read what you get from daily sports betting picks, or check the graded record on the results page.

Frequently Asked Questions

What is a key number in NFL betting?

A key number is a final margin that occurs far more often than the margins around it, because of how football scoring is constructed. Three and seven are the two that matter. In 3,028 NFL games from 2015 through the 2025 playoffs, 14.83 percent finished exactly three points apart and 8.69 percent finished exactly seven apart, which puts 23.5 percent of all games on one of two numbers.

Why do key numbers matter more on a parlay than a single bet?

Because a parlay is a conjunction and every leg has to be right. One leg sitting on a key number is one exposure. Four legs is four exposures, and the ticket dies if any of them lands wrong. In the same sample, at least one leg finished within a half point of the number on 9.35 percent of two-leg tickets and 17.82 percent of four-leg tickets.

How often does an NFL point spread push?

Across all spreads in the sample the push rate was 2.48 percent per game, but that average hides the important part. When the closing spread was exactly three, the push rate was 10.16 percent across 433 games. When it was exactly seven, it was 6.40 percent across 172 games. Half-point spreads cannot push at all.

What happens to a parlay when one leg pushes?

At most sportsbooks the pushed leg is removed and the ticket is recalculated at the remaining legs' odds, so a four-leg parlay with one push pays as a three-leg parlay. That is better than a loss and worse than a win. The detailed rules, including how void and dead-heat situations differ, are covered in our post on what happens when a parlay leg pushes.

Is it worth buying half a point on a parlay leg?

Only when the half point crosses a key number. Moving a favorite from -3.5 to -3 converts roughly a tenth of those games from a loss into a refunded leg, which is real value. Moving from -3 to -2.5 gives that up for very little, and buying through numbers like 8 or 12 purchases nothing, because nothing lives there.

Do totals have key numbers like spreads do?

Not in any comparable way. The final combined score landed exactly on the closing total in 0.92 percent of games and within one point in 6.74 percent, against 7.27 percent of games finishing within half a point of the spread. Scoring totals spread across far more possible outcomes than margins do, so half points on totals are rarely worth paying for.

How does live betting change the key-number problem?

Pregame you accept whatever number the book is showing. In a live market the spread passes through three and seven repeatedly over the course of a game, and the repricing is fastest immediately after a score, when a model is adjusting without a trader watching every snap. That creates moments where the same side is available on the correct side of a key number, which is the opposite of the pregame situation.