The Power of the Key Number in NFL Odds
A half-point changes the treatment of particular final margins. Its value depends on their probability, the attached price, and push rules. Football scoring gives a reason to investigate certain margins; it does not supply their frequencies.
Scoring increments and margin probabilities
The NFL rulebook assigns three points to a field goal and six to a touchdown; a successful kicked try adds one. That makes three and seven useful scoring combinations to consider. It does not establish how often any final margin occurs in a particular sample.
Source: NFL rulebook, scoring.
To estimate a margin distribution, state seasons, competition, regular-season or playoff scope, overtime rules, ties, and quote cutoff. Count the signed margin from the selected team’s perspective. An unconditional frequency of three-point games cannot simply be inserted as the chance that this favorite wins by exactly three.
Price the changed state explicitly
Hypothetical forecast: a selected team wins by four or more with probability 50%, wins by exactly three with probability 10%, and otherwise fails to cover with probability 40%. At -3 priced -110, those states win, push, and lose. Net EV per unit = 0.50 × (100/110) + 0.10 × 0 − 0.40 = +5.4545%.
At -2.5 priced -130, the first two states both win. Net EV = 0.60 × (100/130) − 0.40 = +6.1538%. The expected-return improvement is 0.6993 percentage points of stake under this invented distribution. At -3.5 priced -110, only the first state wins and EV = 0.50 × (100/110) − 0.50 = −4.5455%.
The half-point changes which states win or push, while the price changes what every win pays. A comparison based only on a raw probability gap misses that interaction. There is no universal half-point premium or instruction to buy through three.
Do not infer a subsidy from a stationary line
An operator keeping a spread at three while changing its price does not prove that it is subsidizing the selection. Nor does a model’s mean margin of four establish the probability of covering three. Use the full margin distribution and the settlement terms.
Other scoring combinations can be examined with the same method. Unsupported frequencies for six, ten, fourteen, or seventeen are not used here. If you group historical results into spread buckets, choose them in advance and report the size and uncertainty of each group; one season does not ensure reliable differences.
Continue reading: Understanding Teasers · Implied Probability.
Frequently Asked Questions
Why study margins of three and seven?
They correspond to a field goal and a touchdown plus a kicked try in NFL scoring. Their exact final-margin frequencies require a declared dataset.
Is buying a half-point always worthwhile?
No. Compare complete expected payoffs, including pushes and the price change, under a stated margin distribution.
Does a projected four-point win establish value at -3?
No. A mean margin alone does not provide the probability of a win, push, or loss at that spread.