The NCAA Tournament is consistently one of my favorite sporting events. Nothing else quite matches the sheer volume of games played, oddball stories, on-and-off-the-court emotions, and–most of the time–the spirit of competition unsullied by salaries and fiscal dealings (looking at you, Arizona).
Filling out a bracket started as a cult phenomenon and has exploded into a mass movement. Everyone has a bracket now and everyone is willing to proffer advice on the best ways to make your picks when March rolls around. Alphabetical. Cutest mascot. Regurgitating what you know from the 30 minutes of SportsCenter you watched one time. Each of these methods boasts its own merits, but is there a superior strategy for sealing your selections?
Tim Chartier, professor of mathematics and computer science at Davidson, thinks he’s got some answers. He’s devised a program that allows you to tweak variables and make predictions with software used in predicting trends in finance, retail and health care.
“March Mathness” is constantly undergoing revisions to yield the most successful predictor of tournament outcomes, but, statistically, here’s what Chartier has learned so far:
- Lay off picking 10-seeds and below. Only eight teams ranked seeded 10th or higher have ever won 3+ games.
- If you’re going to pick a Cinderella, look for teams in weaker conferences who have fared well against tougher out-of-conference opponents at the beginning of the season.
- Over 75% of upsets are by 10, 11 or 12-seeds. Most of the time they’re by 12-seeds.
- If you’re picking a first round upset, shoot for team that boasts an offense in at least the top 50–top 30 is even better.
- Every national champion in 14 of the last 15 years has been a 1, 2 or 3-seed. They’ve all come from the top eight conferences and they’ve all been ranked in the top 25.