Mechanisms of Impact of Investor Behavioural Biases on the Performance of Quantitative Investment Models

Abstract

This paper uses literature analysis and theoretical reasoning to examine how investor behavioural biases affect the performance of quantitative investment models. It is believed that overconfidence, herding behaviour and loss aversion, as well as the disposition effect that may result from loss aversion, do not directly affect returns but rather do so through other channels. Specifically, these behavioural biases first change trading behaviour and market conditions, and then modify the meaning of traditional variables such as price, trading volume, volatility and asset correlations, as well as the relationships among these variables and prediction results. That is, overconfidence leads to noisy trading and an underestimation of risk; herding behaviour magnifies market trends but also increases crowded trades and correlations, and the disposition effect can cause investors to delay the recognition of losses and thereby concentrate the realisation of losses over particular periods. Based on the above, this paper adds interaction terms between behavioural states and traditional market variables to the quantitative model to identify changes in market regimes and improve risk control. Based on the above theory, behavioural indicators can be employed as moderators to adjust the effects of traditional signals in various risk scenarios; however, they should not replace price, fundamental and risk data. As this is a theoretical rather than empirical study, the proposed mechanisms have not yet been validated using data from different financial markets.