ABSTRACT
Sports betting markets are an attractive setting to evaluate the degree to which publicly available information is efficiently incorporated into asset prices. Weather information represents one category of potentially undervalued public data, and we assess its impact on wagering outcomes and market behaviour in American college football. We uncover statistically significant effects of weather characteristics on totals market outcomes, indicating that the market does not appropriately assimilate this data into its forecasts. We also demonstrate that weather characteristics are not statistically significant predictors of totals market line movement magnitude. Together, these results suggest the market systematically undervalues weather data. We then develop simple wagering strategies utilizing only three weather characteristics and illustrate the ability to profit.
Acknowledgement
The authors would like to thank Christopher Smith for his assistance with data collection.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 These values are calculated using the listcoef command in Stata version 15.1.
2 Games with no opening line over/under total are excluded. Two observations where Movement is greater than 20 are also removed as these games either contain an obvious release of unexpected information, such key player(s) injury or suspension, or represent clear bookmaker error. The results are robust to changes in this threshold.