Predicting Winners and Evaluating Simple betting Strategies for BCS College Football Games
Abstract
We investigate two sets of questions concerning Bowl Championship Series (BCS) college football games. The first set of questions pertains to predictors for the outcomes of these games. We evaluate whether the differences in rankings between teams and the point spread are good predictors for outcomes of BCS games, which involve top ranked teams in the NCAA Division I Football Bowl Subdivision. We use probit regressions with the differences in ranking, differences in the log rankings, and the point spread as the predictors of the outcomes of the games. The second set of questions pertains to betting market efficiency and testing if simple betting strategies are profitable. We investigate whether betting on the underdog is profitable when the spread is greater than 5 points and also if there are returns to betting the under. We use Z-values and likelihood ratio testing for these analyses.
Keywords: Amateur sports, Rankings, Gambling, Betting, Market efficiency, Prediction markets.
JEL Code: L83