College Basketball

Better Brackets: Before March Madness, Forget About the Top 25 Poll


College basketball’s major conference tournaments get going this weekend, a prelude to Sunday night’s NCAA Tournament selections. It’s not too early to start prepping your bracket strategy—yes, even before the actual brackets are revealed.

To start things off nice and easy, we will focus on picking the Final Four and gauging which of the high-profile numbers attached to tournament contenders—bracket seed and Top 25 ranking—have the most value in making your picks. Later we’ll focus on heavy-quant approaches at places like Kaggle, Power Rank, and Massey Ratings.

Forget those fun No. 14 seed-beats-No. 3 seed upsets in the first round. The key factor that will make or break your bracket is getting the Final Four teams right. Consider this year an approach of picking your four regional champions and then working backward to see how they might arrive there. Surely a 14-seed win might be a fun distraction for a day, but when those Cinderellas wash out, the big money will be made in accurately assessing the contenders for the final weekend.

In terms of portfolio allocation for the Final Four, history suggests staying near the top. Since the tournament expanded to 64 teams in 1985, only three double-digit seeds have ever been among the last four standing.

Based on the historical data, your lowest-risk Final Four lineup should include two No. 1 seeds, a No. 2 seed, and a team from the 3-to-5 range. Focus on finding teams that way. It might prove easier than getting caught up in the round-by-round matchups, which can lead your whole bracket astray. Anything outside this range will get you to the danger zone. Too many top seeds is nearly as dangerous as leaving them all out. The overdogs are the backbone of a winning bracket, with a sprinkling of slightly lower seeds. Less than 10 percent of Final Four teams have been seeded sixth or lower.

What about turning to the experts? The Top 25 rankings start with a preseason assessment, before any games have been played, and continue with weekly updates throughout the college basketball season. From the perspective of picking a tournament champion, it’s almost scary how little the predictive value of these rankings improves over the course of a season, even after the expert voters have watched hundreds of games.

Presumably the Top 25 poll should be better at predicting the Final Four in March than November, but that’s not true at all.

The polls might be fun, giving fans a sense of who is rising or falling, but they are just not that helpful with a bracket. In many years the preseason poll has been better at predicting the Final Four than the poll in the final week of the season. This suggests that the voters often had it right earlier in the year, relying on a loose sense of talent levels rather than a heap of actual game results. Even if any given team lost games during the season, dropping out of the rankings by March, perhaps innate talent comes back to the fore in the tournament and helps fuel a run.

The dismal track record of the Top 25 and the definite value of tournament seeds suggest the selection committee does a good job of who they pick. Trust the seeds when thinking through your four finalists.

Next up this week we’ll focus on more quantitative aspects that can help you better optimize your bracket.

Eric-chemi
Chemi is head of research for Businessweek and Bloomberg TV.

Best LBO Ever
LIMITED-TIME OFFER SUBSCRIBE NOW
 
blog comments powered by Disqus