Abstract
This research examines empirically the effect using broken odds - the rounded off odds reported by horse track operators - on statistical tests of market efficiency in parimutuel gambling. It finds that using rounded-off odds to test for market efficiency, instead of using the exact values, introduces a downward bias in the test statistics for favourite horses. Previous studies, because they relied on the rounded-off odds data, may have overstated the extent of the favourite-longshot bias in parimutuel wagering markets.