Dr. Jay Coleman loves numbers almost as much as he loves basketball.
When the business analytics professor from the University of North Florida stumbled upon a college basketball fan’s website more than 15 years ago, he began a project that would become an annual ritual and garner much national attention every year — particularly during the month of March.
The site Coleman found had data on every NCAA men’s basketball team, which he decided to analyze from a statistical perspective. His topic of choice: predicting the at-large bids for the NCAA Men’s Basketball Tournament.
Determining the teams that make the cut and the reasoning behind these choices is always the subject of much speculation leading up to Selection Sunday in mid-March. Fans on sports talk radio hash out prior injustices and make predictions while waiting for the day to arrive. Coleman’s formula seeks to determine the committee’s behavior based on past data and predict the current year.
Since the first version of the Dance Card, created in 2001 with Allen Lynch, another UNF business professor at the time, there have been two revisions. Those two updates, done with economist Mike DuMond, examine the possibility of bias in selection.
“What we’ve found is that though the committee changes, it is always pretty consistent,” Coleman said. “We’re always close.”
The revisions of the formula have gotten closer. The original version was more than 90 percent accurate. The second and third versions have combined to go 141 out of 146 (97 percent) over the last four seasons.
“We were perfect once,” Coleman said, referencing the 2013 Dance Card, which was an exact match to the NCAA’s final bracket.
Coleman continues to produce an annual Dance Card available at DanceCard.unf.edu, and is now analyzing data from other sports like football, all while serving as UNF’s associate provost for Academic Affairs.
His Dance Card has been featured in Forbes, Business Week, the Wall Street Journal, the New York Times, Sports Illustrated, ESPN the Magazine and more.