Abstract
Previous studies of efficient markets in parimutuel betting isolated only one race characteristic, determining efficiency by comparing subjective to objective probabilities of different groupings. By incorporating regression analysis and looking at a wide range of race specific variables, this study is able to isolate various factors which influence efficiency. Using a data set of 5020 races at 18 US racetracks, a standard favourite–longshot bias was found, which diminishes for races with larger pools and more horses in a field, and increases for races with higher quality fields and maiden races. When track-specific characteristics are factored out, similar results occur and it is also found that races on grass reduce the bias.