I’m continuing my investigation of what we can infer from NFL point spreads. I wrote a little code to compute the conditional probability of the favorite team winning, given the size of the point spread. Here are the results.
As you can see, the game is essentially a tossup until the point spread is at least a fieldgoal. Once the spread exceeds a touchdown, the favorite wins nearly 4/5 times. In the rare instance that the point spread exceeds two touchdowns, then the favorite is a lock to win.
Here’s the raw data used to generate this plot.
spread wins games probwin
============================================
0.00 21 47 44.68
1.00 70 138 50.72
1.50 60 123 48.78
2.00 87 159 54.72
2.50 126 249 50.60
3.00 351 587 59.80
3.50 225 360 62.50
4.00 101 155 65.16
4.50 71 120 59.17
5.00 97 133 72.93
5.50 99 142 69.72
6.00 108 168 64.29
6.50 154 226 68.14
7.00 177 238 74.37
7.50 85 123 69.11
8.00 75 95 78.95
8.50 70 85 82.35
9.00 71 91 78.02
9.50 79 98 80.61
10.00 66 95 69.47
10.50 55 67 82.09
11.00 40 48 83.33
11.50 23 27 85.19
12.00 20 25 80.00
12.50 28 34 82.35
13.00 20 24 83.33
13.50 34 44 77.27
14.00 26 33 78.79
14.50 14 16 87.50
15.00 9 9 100.00
15.50 9 9 100.00
16.00 14 14 100.00
16.50 3 3 100.00
17.00 5 5 100.00
17.50 6 7 85.71
18.00 2 2 100.00
18.50 1 1 100.00
19.00 1 1 100.00
19.50 2 2 100.00
20.00 1 1 100.00
21.00 1 1 100.00
22.00 1 1 100.00
24.00 2 2 100.00
