Abnormal returns in an efficient market? Statistical and economic weak form efficiency of online sports betting in European soccer

Abstract

Online sports betting markets have expanded significantly during the recent decade. At the same time, due to their apparent similarities with traditional financial markets, the academic literature has perceived sports betting markets as a suitable empirical setting for different tests of efficiency.

Employing a data set more extensive than in any previous relevant study, this thesis investigates weak form efficiency of online sports betting in European soccer between seasons 2009 and 2014 from two perspectives: statistical and economic. The statistical tests examine whether subjective probability indicated by average market odds is an unbiased estimator of objective outcome probability. Economic tests, representing the stricter tests of efficiency, inspect whether any betting strategy yields positive returns, utilizing highest odds quoted in the market.

The statistical tests find clear evidence of a persistent favorite-longshot bias and hence of statistical weak form betting market inefficiency. On the aggregate level, both the linear and logit regression models reveal that as the subjective probability of an outcome increases, the objective probability increases more than implied by market efficiency. On the individual odds level, the technique that sorts odds into groups based on subjective probability discovers that the deviations between subjective and objective probability occur on both perimeters of the odds spectrum, not in the middle of it. Thus, in a statistical sense, the betting market appears to be weak form inefficient at high respective low probabilities, while being efficient in between.

The economic tests give a more ambiguous view on weak form efficiency. When considering all the matches in the sample, neither the tobit regression model nor any of the naïve strategies show chances for profitable betting, while some of the odds groups with highest subjective probability demonstrate moderately positive returns. When only including matches with a positive expected value and simulating the associated returns, the sophisticated strategy based on quasi-arbitrage yields no profits, but the strategy that takes into account the favorite-longshot bias generates consistent profits. Therefore, the betting market seems to be inefficient also in an economic sense.

The study concludes that the European online sports betting market in soccer is weak form inefficient. The results are consistent with the earlier literature in terms of statistical efficiency but not in terms of economic efficiency. Due to the lower bookmaker margins in the current market, it is shown that well known statistical biases now also lead to economic inefficiency. The study provides two explanations for the persistence of these inefficiencies: institutional arrangements and market immaturity. Since sports betting markets are not yet as sophisticated as many financial markets, they provide attractive investment opportunities for sharp bettors.