500 Words — Day Forty-Seven: March Madness

For fun, I have developed models to attempt to optimize my odds of winning any given bracket pool and to perhaps enhance the longevity of an untarnished bracket. Unfortunately those models have really been struggling with the first few games in the last couple years, but one model almost won the bracket pool I was in last year, but unfortunately, Gonzaga couldn’t come out on top. Given that this year in college basketball was a little bit closer to normal in terms of scheduling and games played, hopefully this will be a better year for the models, especially given the good amount of parody seen throughout Division I this year.

There are two general strategies that one can take when attempting to win a bracket pool. The first strategy is to be the most right and be conservative with one’s picks and the other strategy is to hope for chaos and hope that nobody else is selecting the type of chaos that overlaps with your bracket. Either strategy requires luck, but if you do it right, the expected performance of any given bracket within the pool should be higher than it would be if you just randomly picked teams or perhaps what is more likely, selecting from your gut feeling.

With strategy one, the goal is to outperform competitors by selecting the perceived best teams in order to get the best possible bracket. The players that typically do the best here are people that know nothing about basketball. Why is this? Because they aren’t reliant on their knowledge of basketball to make decisions. They are more likely to use seeding as a crutch and committee members are much more likely to be good at evaluating teams than your run of the mill college basketball fan. Your normal college basketball fan is more likely to try and be clever and select teams that they think are better versus trust the seed given to them. This tendency leaves them vulnerable to personal bias. Maybe they like one team because they saw a few games were they shot lights out. Maybe they hate another team because they have an obnoxious fan base. Things like that. From my perspective, when I want to attempt to beat chalk brackets, I rely on my models to make the decisions for me. Models aren’t bias free, but it helps to establish the rules before seeing the seedings of the team so I’m not tempted to rely on personal bias.

The second strategy is to hope for chaos. There are several ways to attempt this, but in general, the key is to be contrarian where it makes sense. Now that’s hard given you can’t look at other people’s brackets in the pool, but you do know that everyone is referencing seeding, so you can start there. Also some websites give you information on which teams the population favors. One thing with a contrarian bracket is that you are conceding that you believe that the best teams are vulnerable enough to lose to inferior teams. In a year with bad parody, this may be a bad strategy. In a year with good parody (such as this year), this might be a better strategy. Note that chaos has to break your way. Given chaos is less likely than order, why use such a strategy? Because you are attempting to reduce the amount of competition you are facing. If you pick a contrarian champ, you don’t need the luck to go your way in the early rounds, you just need your champ to survive and advance. With a top team, you might have multiple brackets with the pick meaning more needs to go right in those early rounds. Note there’s a fine line between being contrarian and being clever that we saw fail in the first strategy. Again, using a model to find key matchups that the population is not valuing correctly helps a lot. Also picking the favorite against trendy upsets helps to stabilize a bracket in the early rounds. Picking all 5s in 5–12 matchups is good because conventional advice is to pick one upset there (but the contrarian idea here is that it is very hard to pick those upsets reliably, so let’s not waste any effort there). Even more contrarian would be to advance all 5s to the Sweet 16. Again people like picking upsets in 4–13 matchups and 5–12 matchups. This strategy just bets against that and gives good upside to balance a lower seed champion. Again, not a strategy that necessarily works, but does gives some flexibility while playing against common conventions. This is what being contrarian is all about. It is not about selecting two 14 seeds to win. That is being clever and that often never works. As someone who has picked 14 seeds attempting to be clever in the past, I can confirm that it doesn’t work.

It should be fun to watch my brackets this year. I’ll admit that I have watched close to zero college basketball this year because of how busy I have been, so it’ll be fun. But I noticed that people that don’t watch basketball tend to do better and my models give me information to attempt to make informed decisions without falling into trying to outsmart myself into a losing bracket.