Football NPI Dials
Dials have been set for football as follows:
• Winning percentage / strength of schedule: 25/75
• Home/away win/loss weights: 1.05/0.95
• Quality win bonus threshold: 52.0
• Quality win bonus multiplier: 0.5
• Overtime weight: 100/0
• Minimum retained wins: 5.5
This page will give a short overview of what each one means. For details of how the NPI is calculated, see here; the quick version is that each game is given a value based on outcome and opponent strength, and the NPI is a weighted average of those values.
Winning percentage makes up approximately 25% of a team’s NPI, and the remnant is strength of schedule. The way SOS is calculated in the NPI results in it having a narrower range than the old version of SOS, but it still has a greater influence on NPI than it did in the old 50/50 blend.
The first number refers to the weight of a home loss or road win, the second to a road loss or home win. In effect that means that in a road game, a win counts as 1.05 wins and a loss counts as 0.95 losses. All neutral-site results are weighted at 1.0.
When a team beats an opponent whose NPI is above 52.0, a bonus is applied to that game, equal to (opponent’s NPI minus 52.0) times 0.5. But since this bonus is applied to the value of a single game, it may take more than one such win to have a meaningful effect on the team’s overall NPI. Note that the bonus is not static, but will change over the course of the season along with that opponent’s NPI.
All this means is that five-set wins count exactly the same as three- or four-set wins. (The championships committee overruled the volleyball committee's request to weight five-set wins differently.)
NPI always excludes losses that would raise a team’s overall NPI. With wins that would lower NPI, it’s more complicated; once a team has 5.5 wins (as weighted for home/away), their wins over the weakest opponents are removed from the calculation until one of two things happens (a) they’re left with only 5.5 wins, or (b) all the remaining wins raise their NPI. Timing of these wins is irrelevant to the calculation; the most damaging wins are removed first, regardless of when they occurred. And as with the QWB, it’s entirely possible for it to change, over the course of the season, which wins are damaging enough (or losses good enough) to be excluded.