There Are Methods to Their Madness; Duo Develops Software to Predict Credit Risks or Basketball Pairings

Article excerpt

Byline: ALISON TRINIDAD

Betting which men's college basketball teams will receive at-large invites to the NCAA tournament, aka March Madness, tonight?

Data from two business professors could help.

Jay Coleman, associate dean of the Coggin College of Business at the University of North Florida, and Allen Lynch, associate professor of economics and quantitative methods at Mercer University in Macon, Ga., have devised statistical models that predict which teams will be invited to play the month-long tourney and which teams would win each match.

In the last 13 years, Coleman says, the first model - called the Dance Card - has proven to be accurate about 94 percent of the time; the second - called the Score Card - has been correct about 73 percent of the time.

The professors apply to sports math formulas that also are increasingly used by business forecasters, including managers, investment bankers and actuaries, to improve their bottom lines. Whereas corporate America might use the formulas to determine credit risk for loan applicants or which advertisements are more effective, the Dance Card predicts what the NCAA Tournament Selection Committee will do.

To be sure, Coleman insists that forecasts of any sort are never perfect.

"The forecast is always going to be wrong," he said. "The idea is to be simply better than the other guy, or at least be better than we have been."

Nonetheless, the business of forecasting is getting bigger, if not better.

Morgan Stanley, a large U.S. investment bank, says technology used to analyze and predict business performance is growing about 14 percent a year. Compare that to the general software market, which is growing at a 6 percent tick annually, according to IDC, a leading information technology analyst firm.

That said, it also seems strategies that pad corporate pockets are making their way into the mainstream. For one, there's Numb3rs, a television show on CBS that features a mathematician who uses statistical analysis to help solve crimes. Then there was Moneyball, the 2003 book by Michael Lewis that analyzed how the Oakland Athletics managed a winning record despite having the smallest player payroll among major league baseball teams.

Sports prediction models like Coleman and Lynch's Dance Card and Score Card are the latest pop-culture manifestation of what math and information technology are capable of accomplishing, says Rebecca Wettemann, vice president of Nucleus Research, an independent IT consulting firm based in Massachusetts. …