Favourite–longshot bias
Evidence and explanations#
Documented for decades in horse racing. Snowberg and Wolfers (2010) find it is better explained by misperception of probabilities than by risk-loving preferences. Exchange-traded event contracts show related calibration patterns that vary by category, horizon and liquidity.
How big can it be?#
Small distortions in probability are large distortions in return. In a one-parameter probability-weighting model (Prelec weighting, α = 0.85), a 2% event trades near 4.1¢ and buying it returns about −51% per dollar in expectation, while the 95% side trades near 92¢ and returns about +3% before costs (model and table).
Why it persists#
Correcting an overpriced longshot means buying the expensive side: 96¢ of capital to earn a few cents over the life of the contract. After fees and the cost of capital, the return is often too low to attract enough arbitrage.
Questions#
What is the favourite–longshot bias?
The tendency for longshots to be overpriced and favourites underpriced in betting markets, so bets on longshots lose more on average.
Does it apply to prediction markets?
Related tail distortions are reported in event contracts and vary by category and horizon. Treat tail prices as biased estimates and calibrate them on resolved data.