Extremeness Aversion Bias and Learning Behavior : Evidence from the Sports Betting Market

This study finds that the extremeness aversion bias documented in decision making studies exists in the sports betting market in which real money is at stake. The magnitude of the bias is large that a naive betting strategy utilizing this bias, since the betting market's inception in 2008 through 2012, yields an average after-tax return greater than 47%. The size of the return is more surprising because this betting market provides an unconditional expected return of less than -50% due to low payout rate. I find evidence of learning behavior over time, but not enough to eliminate market inefficiency. The setting of this betting market also allows me to differentiate the two types of extremeness; extremeness based on the positioning of choices versus extremeness based on the statistical ordering of choices. I find that the bias is apparent only when the two types of extremeness are consistent with each other