Link to Google Docs spreadsheet for this post The financial industry isn't the most popular thing in the world these days, but if you're looking for an edge in your NCAA tournament pool, taking a tip from those dastardly hedge-fund managers might pay off.
Slate ran a great story yesterday taking about the problem, and opportunity, in most basketball pools. Generally, the collective wisdom will start to coalesce as the size of a pool grows. In ESPN's bracket contest, which gets millions of entries, you get a great representation of the widsom of the crowd when it comes the tournament (The crowd thinks that North Carolina is the choice to cut down the nets in Detroit).
The crowd, it turns out is pretty damn good at picking the bracket. In 2008, the national bracket finished in the 80th percentile in the ESPN contest. In 2006, it was in the 90th percentile.
But here's the thing -- if you make the same picks as the crowd, you have a very limited chance of winning your pool. Finishing in the 80th or 90th percentile is great, but doesn't win you the pool. If you stick with the crowd, you have little to no chance to score points that other players won't (Imagine a pool filled with identical brackets, to take the point to its illogical conclusion). What you need to do is make smart choices that are contrary to the broader wisdom of the market -- hence, the financial metaphor.
So, how do you do that? I've used the team rankings from the indefatigable Ken Pomeroy, the dean of college basketball stats geeks. Pomeroy's ratings use the concept of offensive and defensive efficiency -- how good they are at scoring each time they have the ball, and how good they are at stopping the other team from scoring on each possession -- to come up with a theoretical winning percentage (Read much more about Ken's ratings at his blog).
The idea is that you have a more objective view of a teams strengths and weaknesses, isolated from all the factors like the quality of their competition, where the games are played, their luck, etc. No statistical system is perfect, but Ken's is pretty darn good, and seems like the best place to start.
Now we can line up the probablity that the KenPom system gives a team to reach a certain round of the tournament with the probablity the crowd has assigned to it. Check the difference between them, and you start to see which teams are over-valued, and which are under-valued.
As I mentioned, the ESPN favorite is North Carolina, which is currently picked to win the national title on 27.9% of all entries. But Pomeroy only assigns the Tar Heels a 9.62% chance of winning the whole thing. That's an asset that's highly inflated.
On the other hand, take Gonzaga. Pomeroy thinks the Zags have a 7.26% chance of winning the tournament, while only 0.6% of ESPN players have picked them. That's an undervalued asset.
Of course, this is a high-risk, high-reward strategy. Stick with the crowd and you probably won't win, although you likely won't finish last. But if that 7% chance of Gonzaga winning the tournament pans out and you're the only one who picked them, you're almost certain to win your pool.
With the risk so low in a pool, there's no reason to play it down the middle -- as Chris Wilson writes in Slate, there's no difference in the payout in most pools between finishing fourth, and four-hundredth.
How do you do it? I've compiled a Google Docs spreadsheet comparing the KenPom numbers with the current ESPN numbers for every team and every round in the tournament. For each round, you'll see the percentage from both sources, and then the delta between them. Teams highlighted in green are undervalued in the ESPN pool compared to the KenPom ratings, and are worth a look (Huge thanks to ESPN and Ken for making all this data available).
I'll be entering a ESPN collective wisdom bracket in its pool, as well as a bracket that's built from this analysis -- we'll see which does better. You can subscribe to my Twitter feed, where I'll post updates throughout the tournament on how they're doing, and I'll post updates here as well.
Like I said, it's high-risk, high-reward. But it's worth a shot.