Learning and loss aversion: evidence from a financial betting market
Item statusRestricted Access
Embargo end date31/12/2100
Ó Briain, Tomás
This research is motivated by a number of open questions in the behavioural finance literature. Firstly, if investors do not learn in a rational Bayesian manner but rather suffer from biases set out in the naïve reinforcement hypothesis, rationality assumptions in individual preference models may not hold. I use a unique longitudinal dataset comprising in excess of 1.5 million fixed-odds financial bets, where bettors perform identical, consecutive decisions which mimic financial choices made in a laboratory, but the use of their own funds departs from the artificiality of an experiment. I present evidence of unwarranted overconfidence generated by reinforcement learning in both real and simulated markets. Secondly, Kahneman and Tversky (1979) state that losses loom larger than gains. I examine whether the disposition to avoid losses is driving behaviour in the losing domain in the dataset and conclude that there is little evidence of loss aversion. I differentiate between betting on Financial Markets, in which agents may perceive an internal locus of control, and betting on the simulated market, where results are uncorrelated and in which the emotions of regret and disappointment may not loom as large. Finally, Odean (1998) provides evidence that investors readily realise paper gains by selling their winning stocks, yet hold on to their losing stocks too long. This loss aversion is consistent with Kahneman and Tversky (1979) prospect theory, however, how long would the investor hold on to a stock that is losing value on a day-to-day basis? Conversely, would an investor rush to sell a stock that has yielded positive returns in each month during the past year? I test the interaction between learning and loss aversion in a financial betting experiment in which two treatment groups are subjected to consecutive gains or losses.