By Ray Downs
By Lindsay Toler
By Danny Wicentowski
By Lindsay Toler
By RFT Staff
By Lindsay Toler
By Allison Babka
By Lindsay Toler
Spring break finds Saint Louis University math professors Bryan Clair and David Letscher hunkered in front of a dozen television sets at Flannery's Pub downtown. It's the first round of the NCAA basketball tournament, and the academics can ill afford to miss a moment of March Madness.
Their reputations are on the line.
Earlier in the week, the duo published an article in the New York Times outlining a mathematical strategy for winning the thousands of Internet and office pools that take place this time of year. As even a novice bracketologist knows, the outcome of these first-round games can flush one's pool future down the commode. Even more so for Clair and Letscher, who specialize in playing against the percentages.
"Oh, God," Clair cringes as the University of Oklahoma's Sooners fall to the Panthers of UW-Milwaukee. "We're going to do terrible!"
In the 72 hours since the duo's article hit the pages of the Times, 50,000 people have logged onto Letscher's SLU Web site (dehn.slu.edu/sports) to eyeball the professors' predictions, the product of a year and a half of formula-honing.
Rather than forecast the outcome of the tournament based solely on the track records of the teams involved, Clair and Letscher amass data from the Internet in order to discern which teams the masses favor. These are the teams they don't want in their Final Four especially if they're consensus picks in massive online pools such as those operated by Yahoo and ESPN, in which the number of entries reaches into the millions.
The professors feed their data into their complex software program, which combines basketball stats with public sentiment and voilà! brackets that look like few others.
Clair and Letscher output three different ostensible pool entries: a version customized for small "office" pools of 50 participants, one for middling pools of 5,000 and one for pools of 500,000 to post in mega-competitions like ESPN's. While most of the nation believes either Duke or Connecticut will snag the national title on April 3, all three of Clair and Letscher's brackets have Texas topping Florida for all the marbles. (Close to home, they had Southern Illinois University's Salukis going down in the first round to West Virginia in all three versions.)
"Our logic rewards the long shots," explains the bespectacled Letscher, who seemed quick to shrug off early first-round losses of Oklahoma and Marquette. (Prior to the tourney, Vegas oddsmakers had UConn at 3-1 and Duke at 7-2; Texas was 12-1 and Florida 20-1.) "The misconception is that we've developed a formula to win the tournament. That's not true."
Rather, Letscher contends, he and Clair have found a way to identify "value" bets in much the same way a savvy horseplayer hones strategies for wagering on Thoroughbred racing.
Boasts Letscher: "We believe we've upped the odds a thousand times."
In other words, in a betting pool that contains 1 million contestants, Clair and Letscher believe they have a 1 in 1,000 chance of winning. By the same token, this means their chances of winning the pool this year or anytime soon are slim.
"We may not see our formula pay off in our lifetime," Clair concedes. "Probably 80 percent of the time we'll score in the bottom 20 percent."
The professors, both in their thirties, hit on their theory one afternoon in 2003 while discussing the NCAA tournament. That was the year 51 percent of the participants in ESPN's online pool picked Kentucky to win the national championship. The Wildcats got only as far as the Elite Eight.
"Kentucky was a good team," Letscher recalls. "But there's no way any team entering the tournament has a 50-50 chance of winning the championship."
With that in mind, the mathematicians set about decoding March Madness. Their paper, "Optimal Strategies for Sports Betting Pools," is now making the rounds of academic peer review.
Bob Goetz, weekend sports editor for the New York Times, came across the document while trolling the Internet in early March.
"I was looking for professor types who could talk about the bracket as a math problem," says Goetz. "Other people have done similar studies, but I liked theirs because they incorporate this huge body of information out there on the Web that wasn't available for earlier studies."
After following up with the professors by phone, Goetz commissioned them to boil their theory down to an article, which he published in the Times' March 13 sports section as part of a package titled "How to Win Your Pool."
The editor's take on Blair and Letscher's first foray into journalism?
"They were terrific. I'm not going to tell you I understand the math behind their work, but their conclusions are very understandable," says Goetz. "It was one of the most e-mailed articles of the day."
Of course, geeks into gaming is nothing new. In 1962 mathematics professor Edward Thorp published the seminal Beat the Dealer,in which he applied a mathematical system of card counting to win at blackjack. More recently, in the early 1990s six Massachusetts Institute of Technology students famously applied a variation of Thorp's theory and came away from Las Vegas casinos with millions of dollars. They were caught only when turncoats within the posse sold names and photos of their colleagues to the casinos' in-house security.