Management Science, 2018, Vol 64 (4), pp. 1957–1973
We use novel data on individual activity in a sports betting market, to study the effect of past performance sequences on individual behavior in a real market. The idiosyncratic nature of risk in this market and the revelation of assets' true terminal values enables us to disentangle whether behavior is caused by sentiment or by superior information about market mispricings, and to cleanly test two prominent theories of momentum and reversals — the regime-shifting model of Barberis, Shleifer, and Vishny (1998) and the gambler's/hot-hand fallacy model of Rabin (2002). Furthermore, our long panel enables us to study the prevalence across individuals of each type of behavior. We find that i) three quarters of individuals exhibit trend-chasing behavior; ii) seven times as many individuals exhibit behavior consistent with Barberis, Shleifer, and Vishny (1998) as exhibit behavior consistent with Rabin (2002); and iii) no individuals earn superior returns from momentum trading.
Forthcoming at the Journal of Finance (Replications & Corrigenda)
Barras, Scaillet, Wermers (2010) propose the False Discovery Rate to separate skill (alpha) from luck in fund performance. Using simulations with parameters informed by the data, we find that this methodology is overly conservative and underestimates the proportion of nonzero-alpha funds. E.g., 65% of funds with economically large alphas of ±2% are misclassified as zero-alpha. This bias arises from the low signal-to-noise ratio in fund returns and the consequent low statistical power. Our results raise concerns regarding the FDR’s applicability in performance evaluation and other domains with low power, and can materially change its conclusion that most funds have zero alpha.
Revise and Resubmit at the Review of Financial Studies
Using trading data from a sports-wagering market, we estimate individual dynamic risk preferences within the prospect-theory paradigm. This market's experimental-like features greatly facilitate preference estimation, while our long panel enables us to study whether preferences vary across individuals and depend on earlier outcomes. Our estimates extend support for existing experimental findings — mild
utility curvature, moderate loss aversion, probability overweighting of extreme outcomes — to a real financial market, but also reveal that risk attitude is heterogeneous and history-dependent. Applying our estimates to a portfolio-choice problem, we show prospect theory can better explain the prevalence of the disposition effect than previously thought.
We propose a novel methodology that jointly estimates the proportions of skilled/unskilled funds and the cross-sectional distribution of skill in the mutual fund industry. We model this distribution as a three-component mixture of a point mass at zero and two components - one negative, one positive - that we estimate semi-parametrically. This generalizes previous approaches and enables information-sharing across funds in a data-driven manner. We find that the skill distribution is non-normal (asymmetric and fat-tailed). Furthermore, while the majority of funds have negative alpha, a substantial 13% generate positive alpha. Our approach improves out-of-sample portfolio performance and significantly alters asset allocation decisions.
We develop a structural model of behavior that accounts for individual heterogeneity within and across utility ‘types’ characterized by different features of risk preferences, and we estimate it using a unique dataset of individual activity in a sports wagering market. In particular, we estimate the population distribution of utility curvature, loss aversion, and probability weighting, and we evaluate their relative importance in explaining behavior. We find that prospect theory fits behavior in this market, with parameter estimates that are broadly in line with those estimated in the lab, and with substantial heterogeneity: on average, individuals are risk averse over gains and risk-loving over losses, exhibit some loss aversion, and overweight the probabilities of extreme outcomes. Furthermore, we find that utility curvature alone does not adequately explain individuals’ choices and that, while loss aversion is important, probability weighting is the most prevalent behavioral feature of risk attitudes: Two thirds of individuals exhibit loss aversion, but almost
all exhibit probability weighting.