Market Efficiency in Racetrack Betting

Abstract

Betting at racetracks and investing in the stock have several characteristics in common. In neither situation are future earnings known with certainty. Both situations are characterized by a large number of participants and by the availability of extensive information and professional advice. Horse racing data, therefore, permit interesting tests of the efficient theory. Numerous authors have analyzed horse racing data and various conclusions have been confirmed: (1) There is a systematic tendency to overbet long-shots and underbet favorites (Rosett 1965, 1971; Snyder 1978; and Ali 1979). (2) The tendency to overbet long-shots is strongest in late races (McGlothlin 1956; Asch, Malkiel, and Quandt 1982). (3) The utility function of bettors is convex, indicating risk love (Weitzman 1965; Asch et al. 1982). Figlewsky (1979) tested hypotheses concerning whether the betting market is efficient with respect to track odds and handicappers' picks of winning horses. By fitting a logit model with horses' winloss records representing the dependent variable and track odds and handicappers' picks the independent variables, it is shown by likelihood ratio Do inefficiencies in racetrack betting permit profitable strategies to be devised on the basis of observed betting patterns? Logit analysis of data indicates that profits cannot be earned in win betting, although it is possible to outperform the average bettor substantially. Surprisingly, profits net of transactions costs on place and show betting appear possible, although there are reasons to suspect that such opportunities may not be exploitable on a substantial scale. The findings do not suggest irrational behavior by bettors.