AN ALTERNATIVE EXPLANATION OF THE FAVORITE-LONGSHOT BIAS

Abstract

Empirical studies of horse race betting in the US, the UK, Australia, and Germany have empirically established the so called favorite-longshot bias. It was found that bets on longshots on average lose much more than bets on favorites. The theoretical literature on wagering markets has offered a variety of explanations for that bias. One of the most prominent is the assumption of a homogeneous bettor population with a preference for risk. However, the risk-love explanation has also been severely challenged. We add to this challenge by proposing a different explanation of the favorite-longshot bias. We show that if populations of bettors have only noisy estimates of horses’ true winning probabilities, a favorite-longshot bias will be the market equilibrium outcome even with risk neutral bettors and even if the median estimate is correct. We provide evidence on four different types of bets broadly consistent with the noisy estimates assumption but not with the risk-love explanation.