Sunday, March 15, 2009

Bracket Lesson #3: "Clutch" Vs "Overrated"

Probably the most controversial bracket lesson that I have is actually the most successful. Just look at the teams I picked to overachieve and underachieve last season, and it's remarkable how accurate that list was. I picked a 4, 5 and 9 seed to underachieve, and all three flamed out in the first round. Of the teams I picked to overachieve, I picked an 11 seed that made the second round, a 7 seed that made the Sweet Sixteen, and a 6 seed that made the second round. So the "overachieve" teams all did at least as well as their seed would suggest, and none of the "underachieve" teams did better than their seed would suggest.

So what's the point?

The point is that the college basketball season is a small sample size. And while you can get some anomalies over short periods of time, over the long time teams revert to the mean. The mistake that people make is that they see a team that was so clutch during the regular season, and project an even better performance during the postseason. First of all, teams that won a lot of close games were probably some combination of "clutch" and "lucky". Second of all, even if there was no luck involved, they still played so clutch during the regular season that even a very clutch performance in the NCAA Tournament would only cause them to "achieve"... they're not going to shock anybody as a Cinderella. Whatever clutch play they had is already built into their seed. And over the long run, teams will always revert to the mean.

In other words: pick AGAINST teams that won a lot of close games, and bet on teams that lost a lot of close games. This might seem counter-intuitive, but it's a well known issue for those in the statistical community. We see it a lot in Major League Baseball, where statisticians talk about a team's Pythagorean record, and where you can bet the house that a team that plays well ahead of its Pythagorean record is going to come down to Earth in the second half of the season. Teams with great closers and clutch hitters will finish ahead of their Pythagorean, as you'd expect, but never by more than a few games.

So since I'm sure some people just want to know which teams they should keep an eye on this Tournament, I'll just list here some teams that I feel should be expected to over- or under-perform. If you're wondering where I came up with this list, it's mostly just a study of the Sagarin ratings, as well as my own intuition from watching teams this season.

Likely to overachieve:
Arizona State
West Virginia

Likely to underachieve:
Florida State
Wake Forest

No comments: