This is Tech

The Science of March Madness:

Game Theory applied to your Office NCAA Tournament Pool

At This is Tech we like sports, but what we really like is the technology behind sports. With that in mind, we bring you the ThisIsTech.com primer on NCAA tournament pool strategy followed by an overview of game theory as applied to tourney picks from Professor John Duffy of the Experimental Economics Laboratory at the University of Pittsburgh.

As a non-mathematician and newcomer to game theory, my Final Four pool strategy and advice is loosely based on the following principles:

1. The Bigger the Pool the Crazier the Picks. If your office pool is large, you aren’t going to win if you pick all favorites, and even if you do you are more likely to split the pot with someone or otherwise tie.
2. Betting Against the Crowd can Outweigh Betting of the Favorite. In rounds where similarly seated teams play each other, there can be a slight advantage to choosing the worse seated team. The reasoning here is that a higher percentage of players are going to pick the favorite, but the difference in the abilities of the teams is often small. The advantage of betting against the crowd can to outweigh the disadvantage of betting on the worse seated team.
3. Capitalize on Irrationality. All things equal, you should bet against your home team. Other players don’t act rationally when loyalty to their home team is a factor. Betting against your home team can be an opportunity to bet against the crowd without betting on an underdog.
4. Save Riskier Picks for Later Rounds. Early loses don’t just hurt initially, but eliminate the possibility of win points in subsequent rounds that your chosen team can’t play in. As the number of remaining rounds decreases, picking a loser has a less lasting effect. This factor depends to some degree on how the pool is scored.

Now for the big guns. Professor John Duffy is a Ph.D in Economics and teaches Introduction to Game Theory at the University of Pittsburgh (Seated #4). Professor Duffy attended my alma mater, the University of California, Berkeley (not in the tournament), and got his Ph.D. from UCLA (Seated #1).

Full Audio Interview Below: (hit play to listen)

Excerpt Interview Transcript:

This is Tech: What is game theory?
Professor Duffy: Game theory is a branch of applied mathematics that’s looking at strategic interactions among individuals, participating in strategic settings where the outcome for an individual depends not only on his own choices, but also on the choices of others. Many strategic settings, including March Madness, also involved some element of chance and game theory, and take that into account. Standard game theory assumes that players are infinitely rational, but the more recent development of behavioral game theory takes into account biases and heuristics in individual decision making and it might be possible to use some insights from behavioral game theory for March madness picks.
This is Tech: So, is there a classification for this type of game in terms of game theory?
Professor Duffy: Yes, this is essentially a 6-round single elimination tournament. There’s essentially 64 teams that are playing a round, and with 64 teams, and you got to pick 63 games. All but one of the 64 teams loses once. Sixty three games, that’s a lot of games to pick. So it’s a tournament where you got to make 63 choices, there’s element of chance at every step. And you got to consider not only chance, and the behavior of other individuals in your pool, but you’ve got to consider the fundamentals here which are the seatings of the teams.
This is Tech: As far as the advice that behavioral game theory can provide, what are, what are some things people should be thinking about?
Professor Duffy: First, as a disclaimer, when you think of strategies it’s, it’s probably useful to invoke the words of William Goldwyn who was, writing about the predictions of experts in terms of movie box office success. He wrote famously that “nobody knows anything.” And so you have to start with the premise that even the guy in your office who claims to know the most, when it comes to March madness, all bets are off. But, but, let’s ask in a second what’s your objective. Um, is it to win the office pool or is it not to look too bad relative to the know-it-all guy who’s running the pool? If I knew how to win the whole thing, I’d be rich and not have the time or interest in giving free advice on podcasts. But if it’s the second objective of not looking too bad or, or doing it as best as you possibly can, I can offer you some free advice, but you know, consider the price. So, the initial seating of the teams is, the best indicator of team quality available for the start of the tournament, and probably the rational approach or the traditional game theory would be simply to just pick the seeds all the way through. That’s certainly the simplest strategy. Go with the higher seeded team and, as far as I’m aware of that no one has devised any strategy that has successfully beat, go with the higher seeded team as a strategy, consistently. Of course, somebody in your pool will beat it, but no one has found an algorithm or strategy that will consistently do that. Another related strategy is to use predictions derived from Las Vegas odds, you can look at the point spread. These are readily available on the web or what I like are prediction market such as tradesports.com which has a market going now for the sweet 16 and for the championship winner. And there you can look at the trade price that gives the view of many market participants as to who’s most likely to be in the sweet 16 and the championship game. But both of these are subject to biases, both the Las Vegas odds or the trade sport prices, but generally, these are the best predictors of success you can obtain. And you’ll want to consult these resources, at the latest possible date before submitting your predictions as the information can change. Now, that’s just relying on fundamentals. And that’s a more traditional game theoretic view. To think about behavioral game theory, you might want to consider two things. One is the size of the office pool. If it’s small, then, then that’s where game theory maybe has more of a role to play as the pool gets larger, it’s going to be less effective. But one of the insights of behavioral game theory is, sometimes called step level or K-level reasoning which is the notion that individuals form beliefs about other people’s beliefs but to a more limited degree than, than in infinitely rational game theory. For instance, you might know that there’s a bias in favor of a home team or you might know that ah, some colleague is a big Notre Dame fan because, ah, they went there. You may be able to exploit this fact by reasoning that they’re going to pick, say Notre Dame to go to the final four, not rational in terms of seeding for this tournament. And so you can bet against them all the way or some of the way. That’s a, that’s a way of using your knowledge about the beliefs of others to gain some advantage. And there are some people that will practice even higher degrees of reasoning than that. One strategy and, and you mentioned this before we start talking is, betting against your home team. This can be thought of by lots of people and so that’s a sort of first level depth of reasoning over a sort of zero level of emotionally following the home team. You can move to a second order level of reasoning which is to try to think more rationally about how the home team is going to do and there again you might rely on the seeds. Those are some insights aside from, so the traditional view has followed the fundamentals and the behavioral view is to exploit the biases or heuristics of people that you know to be in the office pool.
This is Tech: So who are your picks ?
Professor Duffy: Well you know, from a mathematical perspective, I’d go with the top seeds, so North Carolina looks ah, unstoppable and ah, I also like UCLA. So, I would say for the final, North Carolina and UCLA. Unlikely, we’re going to get ah, both of those teams possibly, any of them, but that’s what you’d have to go with, the seeds the way they are. I also like Kansas and, and, and Stanford, surprisingly is my fourth pick for the final four. So that’s a number 3 seed. I like them anyway. [Interestingly,] Memphis seems to have the lowest current trade price on trade sports of the top seated teams.
This it Tech: Thanks Professor Duffy!

Join the ThisIsTech pool on yahoo.com and go head to head with me (Mike Schneider) and Professor Duffy. No entry fee, no prizes, just an opportunity to see how you stack up against your tech peers. Keep in mind that true to game theory, if everyone uses the strategies described in this article, they wont work.