I found some articles having to do with strategies for choosing the NCAA basketball tournament brackets. In particular, I thought this article was pretty good. The article basically states that a contrarian strategy is a solid strategy if all you care about is making money on pool entries in the long run. With the contrarian strategy, you find teams which have a higher chance of winning and moving into later rounds than most people give them credit for and choose them. Notice, this doesn’t imply that they are favorites, even by any statistic, but as long as most people will choose against them, there will be a high payoff if they do win, since you will move to the top of your pool.
This makes the most difference when choosing teams that will make a win over a favorite and move on the top the Final Four. For an example from this year, Gonzaga will be playing UNC in the sweet sixteen. Many people have UNC winning the entire tournament and a large percentage of entries on ESPN have UNC making it to the final four. However, Gonzaga has a higher chance than those numbers suggest. Based on this fact, choosing Gonzaga over UNC will offer a huge advantage if Gonzaga does win since all those people just lost the opportunity to make points in the later rounds, where each game is worth more points (each game is worth twice the amount of points they were worth than the previous round, ensuring that in every round there is 320 possible points, assuming that the first round games were worth 10).
One of the interesting parts of this approach is that it completely contradicts the basic wisdom I was taught, which was to randomly choose some upsets in the first rounds, but then to go with who you thought the winner was going to be in the later rounds. This advice makes sense because of the relatively small information we usually know about the more obscure teams in the 8-9, 7-10 games, etc compared to the fact that we usually recognize the names of the #1 and #2 seeds and might even know some of their stats. But if we assume that everyone else in our pool follows that strategy, then the winner out of them would be largely based on the random happenings of the first round games, because their later round brackets would be too similar for one person to gain any significant point advantage. We also did this since we knew that losing a Final Four team almost guaranteed a near-last place finishing. But in reality, if you’re not in the money, does it really matter where you were? Knowing that, you would probably be better off simply betting the better seeded teams in the early rounds and then starting to make some “upsets” in the sweet sixteen when the differences between seeded teams will be smaller. By doing this, we increase the likelihood and the relative values of our upsets.
Looking at it this briefly with some math, choosing Gonzaga to win their division and make the Final four is worth total of 150 points (10 for their first round win, then 20 for the next, then 40, then 80). If they do make it, then against someone who chose UNC to make it to the Final Four (which most people are betting on), you would have a 120 point advantage, which is the equivalent of calling 12 first round upsets correctly. So basically, if you are looking to make some ridiculous calls and go against the grain, it makes more sense to call a later round game than an earlier round game. But it is important to note that this situation only applies when you have good reason to believe that the popular chances of the underdog winning are lower than they really are, which goes back to the root of problem of estimating probabilities of teams winning.
In the article I linked to above, there is another link to an academic paper that someone published on the matter and there was some other discussion on the blog of the guy who wrote Freakonomics. Two things there…
1. When I was looking early for literature on NCAA strategies, I should have known to look to economists rather than mathematicians since economists are much more likely to be cool and care about this type of thing.
2. All this literature illustrates one of the reasons that I want to remain around thinkers for the rest of my life; Even common events become extremely interesting when approached in an intelligent manner and reading the ideas of others just makes it one thousand times more interesting.
p.s. Apparently the problem of determing whether or not you still have a chance in the NCAA pool is NP-complete if we don’t fix the number of teams that enter it. Pretty cool.
