Super Bowl mismatch note of the day

I mentioned point differential the other day. I used ratio rather than raw differential (where the Patriots/Giants mismatch is also the greatest in Super Bowl history) because league context makes a difference. It's easier to outscore an opponent by 100 in a league where the league averages 400 points per team than in one where the league averages 200 points per team.

But there are other things that you can do with point differential. Or, at least, with points scored and allowed. Bill James did some work with baseball, and discovered that most teams winning percentage ends up being very close to what he dubbed the "Pythagorean" winning percentage. That is, the winning percentage will be very close, in most cases, to the ratio of the square of the points scored to the sum of the squares of points scored and points allowed. P^2 / (P^2 + PA^2). Work with other sports shows that the relationship holds pretty well. In football, people have found that the best exponent is 2.37.

So we look at Super Bowl teams. In nine of the previous games, one team had a Pythagorean winning percentage that was 24% or more better than the other. Those nine teams are 9-0. There have been 13 games where one team had a 20% or more advantage, and the team with the advantage is 10-3. The two biggest upsets (1980, where Philadelphia's Pythagorean winning percentage was 23.5% better than Oakland's and 2006, where Chicago's was 22.4% better than Indiapolis') occurred when the team with the lower Pythagorean came from a dominant conference. In 1980, the AFC was 33-19 (.634) vs. the NFC. In 2006, the AFC was 40-24 (.625) vs. the NFC. The third best Pythagorean advantage team to lose was the 1968 Baltimore Colts.

The biggest Pythagorean advantage in the first 41 Super Bowls came in 1979, when the Steelers Pythagorean winning percentage (.749) was 42.4% better than Los Angeles' (.526).

The New England Patriots Pythagorean winning percentage (.860) was 60.4% better than the Giants' (.536).