Estimation of expected return integrating real-time asset prices implied information and historical data

Abstract

In this paper, we develop a novel estimation for expected stock returns combining forward-looking information implied by real-time asset prices and backward-looking information implied by historical data. Considering a general heterogeneous market composed of both informed investors and noise investors, we investigate the market equilibrium characterized by the expected returns, risk-neutral moments and market portfolio. To mitigate the negative impact of the market noise on the forward-looking information implied in market equilibrium, we then incorporate historical data and propose the combined estimation for expected return within a Bayesian framework. The combined estimation is adaptive to the market composition and adjustable to changes in market states. Monte Carlo simulations and empirical studies are performed to validate the merits of the proposed approach.