Wednesday, February 1, 2012

Super Bowl Pick

0-1 last week. Turns out recovering fumbled punts is about as effective as hitting Hail Mary's. Not that we're bitter. Sometimes the breaks don't go your way. But we are still short the Giants. We've been wrong or unlucky in that position for weeks now but are going to hold to it.

NE (-3) vs NYG [MP= -6.13]

That's a 3.13-point edge, putting it squarely in the "Other Plays" category. And gives us a little room for liking the Pats even if you grant that the Giants are better than they were in the regular season. And we grant less of that than most people. Over the last 12 years, there is very little movement in the playoffs from a team's regular-season performance. There is SLIGHT regression to the mean, but actually suprisingly slight. And performance at the end of the year -- whether last 1, 2, 3 or 4 games -- doesn't predict anything at all.

Of course there is a lot of noise in playoff performance -- regular season performance is far from deterministic. We just find that our regular season #s go through more or less in full, and that late-season trends don't matter. There's just a bit of research out there that suggests people will "see" it otherwise.

Are the Giants an exception? Entirely possible. Certainly out of those last 144 playoff teams there have been some that got better over time. As there have been some that got worse. It's just that the trend generally doesn't matter, so you're favoring some bits of evidence here and there over the base rate. Again, there is years of research on the tendency to do that and how well it works out (in short: it doesn't).

With a 3-point edge you have to be VERY sure of those bits to like the Giants. We're going with the base rates.