Abstract
In horse racing, “favorite–longshot bias” is a well-known phenomenon around the world. This bias indicates that people tend to bet their money excessively on the horses with low probabilities to win. According to prospect theory, people overbet those horses because they overweight their winning. We present empirical evidences suggesting that this favorite-longshot bias exists in the trifecta market, the most popular betting system in Japanese horse racing. Comparing with past research about a horse racing, our data is more comprehensive in that we cover all betting data including loser's tickets and in that we directly analyze an actual number of votes not estimating voting number from odds. This paper extends the area of empirical researches in behavioral economics by utilizing a huge data set which describes actual behaviors of economic agents towards extremely low probabilities.
