When Davidson math and computer science professor Tim Chartier and his students aren’t helping coach Bob McKillop and the Wildcats’ men’s basketball teams scout opponents, they are busy tweaking and testing methods that assist college basketball fans in their quest to predict the elusive perfect NCAA tournament bracket.

This time of year, Chartier is in demand. Using a data tool devised by Tresata, a Charlotte-based big data company that’s been successful in finance, retail and health care, he is exploring even more possibilities.

A person can spend hours on their “March Mathness” website, playing with variables such as how much weight to give a road win or whether games played later in the season are more predictive than those played in the opening weeks.

Literally, hours.


Bracket busted? Build a
streak of correct picks. 

In the days leading up to Selection Sunday, he’ll prepare the data for his students to analyze, meet multiple interview requests from various media and field questions via random emails from high school students or retirees who are determined to win their March Madness bracket challenge pool this year.

But he’s not going to tell anyone how to fill out their bracket.

“If you really look carefully at what I do, I don’t actually tell you, I direct you to it, but you have to make the mathematical modeling decisions to create your own bracket,” Chartier said. “I like to tell people, I’m not going to do your homework for you. But I do give you a really great tool to help you play potentially quite well.”

But the methods do work. Chartier’s students have finished in the 99th percentile of the ESPN Tournament Challenge.

For the past year or so, Chartier has been trying to improve one aspect of his system, working on a model that determines how teams fare in specific matchups as opposed to selecting one team simply because it’s rated higher.  

One of his students, Max Schimanski, is working on a model that better identifies potential giant killers or Cinderellas. Another found little difference in predictive quality when using a team’s total points scored, compared to using points per possession.

Once the bracket is released, the radio interviews begin for Chartier.

“Sometimes I’m more of a talking head for the work that we do,” Chartier said. “We’ve worked out what they’ve done, I know how they’re doing it. But the results and the talking points are what they’re given me.”

Here are five tips gleaned from Chartier’s team looking at all regular season data dating back to 2002 (a 9 seed defeating an 8 seed is not considered an upset):

1. Want to pick a team with a seed of 10 or higher? Keep these stats in mind: Of teams with a 10 or higher seed, only two teams have won four games in the tournament (2.3 percent) and only four teams have ever won three games (4.5 percent). Of these teams, no team was higher than a 12 seed.

2. For teams in weaker conferences (conference RPI greater than 10), it is more difficult to tell how well they will play against stronger teams. To get a better sense of their strength as a team, look at the out of conference games that they play at the beginning of the season. Even if they lose, if it's a close game or they limit the number of points scored by the other teams, that may indicate that they are a potential cinderella team. 

3. Seed Stats: 76 percent of upsets are by 10, 11, or 12 seeds (27 percent by 12 seeds alone)

4. One-third of lower-ranked teams who win in the first round are ranked within the top 30 offensively -- 55 percent were ranked within the top 50.

5. Stats of winning teams  - in the past 14 years, every national champion except one was a 1, 2, 3 seed. The exception was Connecticut, a 7 seed. Every winner has been within the top eight best-or-strongest conferences. Every winner has been ranked within the top 25 (using KenPom.com pythagorean ranking method).