Two behavioral economists at the Wharton School have recently published a statistical analysis of win-loss patterns in NCAA basketball games. Jonah Berger and Devin Pope collected data from more than 6500 games since 2005, and guided by a psychological/ economic model called Prospect Theory, they cranked their win-loss data through a specialized regression analysis (probably not a standard offering in Microsoft Excel statistics). Then they replicated this modeling with data from a lab experiment using keyboard striking instead of basketball moves.
Their primary finding from both datasets? Losing can be a powerful motivator. Both basketball teams and keyboard strikers were apt to move from a position of slight disadvantage to winning. In the case of the basketball scores, the teams that were one point behind at halftime were more apt to win than teams that were one point ahead at halftime.
Because Prospect Theory addresses the problem of decision making under uncertainty–in fact, decision making in which all the alternatives involve uncertainty–I began to wonder if March Madness coaches could use Monte Carlo simulation to strengthen their strategy. Maybe the coaches could use the same NCAA data and Berger and Pope’s results to run various point-up, point-down scenarios through their Monte Carlo software. This would allow them to anticipate their strategies to respond to any number of halftime scenarios.
This kind of risk analysis could certainly rationalize game strategy. But then March would be March without the Madness, wouldn’t it?