ABSTRACT
In this research we analyze more than 13,000 new races run at eighteen Japanese horse tracks. We examine the relationship between breakage (the rounding down of pay-outs to winning wagers), betting turnover (the dollar amounts bet), and betting market efficiency. The evidence across Japanese horse tracks indicates that tracks with high turnovers are more informationally efficient than tracks with low turnovers. We also find that breakage costs are systematically related to betting market efficiency. We investigate the possibility that bettors have preferences over the skewness of betting returns in addition to their level and variance, and we relate this to betting turnover as well. The new evidence leads us to reject the skewness-preference model at tracks with a high volume of betting; however, the skewness-preference model is consistent with betting behavior at tracks with low betting turnovers.
