How good are your weekly picks?
- November 1, 2011
- UCI cognitive scientist Michael Lee studies science behind sports predictions
For those who follow football (as if there was anything else on TV through fall), late October upsets for Oklahoma and Wisconsin meant a mid-season shake up in the college BCS rankings, and some presumably upset Vegas odds-makers. Following the action both on the field and in the over/unders (both by observation only) has been UCI cognitive sciences professor Michael Lee, one of football’s newer fans. The Aussie native traded in his love for Australian Rules with the closely matched American game in 2006 when he joined the UCI faculty. In five short years, he’s become somewhat of a regular at Bruins games, and for his professional picks, he’s earned several Fantasy Football trophies that can be seen proudly on display in his department chair office.
Yet for Lee, who is, after all, a scientist and information junky, the most intriguing part of the game lies not in the on-field action, but in the behind-the-scenes work that goes into making accurate weekly picks.
“Making picks in college football means going through a large number of games each week, and making decisions on which team will win each game,” he explains. “For some match ups, you will search through a lot of information to try and make a good prediction; for others, you’ll feel like you need to do a lot less work. You don’t get feedback until the games have actually been played, so how do you work out how thoroughly you look for information and where you stop?”
While the professor is quick to state that he doesn't gamble on the games, the general information issue is at the heart of his newly funded $222,600 Air Force Office of Research study.
Using computer simulated experiments, Lee is collecting data on how people gather information, and to what degree they use it, to make decisions in dynamic, changing environments.
“Many existing models of human decision making say relatively little about the search process, and even less about how search is self-regulated as the nature of available information changes, and the goals and circumstances of the decision-maker shift,” Lee says. He hopes to address those issues by extending existing sequential sampling models of human decision-making which would take these factors into account. The resulting model could have wide applications, he explains.
“If we know when and why people change how much information they gather before making a decision, we could adjust how much information we present to decision-makers so as not to overload or under prepare them,” he says.
In addition, since the model will be implemented as a computer program, he says some decision making processes could potentially be automated to emulate what a human would do, but without fatigue, the need for a salary, or, in some cases, placing them in danger, the latter of which may be of keen interest to the study’s funding agency.
For sports fans, Lee’s model could mean fewer upsets and a greater degree of confidence in weekly picks, if presented with the right information. Perhaps not in time for football this year, but basketball is right around the corner…
-Heather Wuebker, Social Sciences Communications
-photo ©istockphoto.com/Adam Kazmierski