In Search of Models for Stock Return Forecast

• Activity Number: 337
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Shaobo Li*+
• Keywords: Forecast ; Dynamic Linear Model ; MCMC ; Model Averaging

Abstract

We attempt to investigate forecastability of US stock market return by using a set of information. Given the high dynamicity of stock return, we propose to develop Bayesian Dynamic Linear Model (DLMs), which assumes model coefficients to be time-varying and has high flexibility in model specification. Markov Chain Monte Carlo (MCMC) Bayesian method and Maximum Likelihood Estimate (MLE) approximated by Quasi-Newton nonlinear optimization algorithm are used for DLMs parameter estimation. Model Averaging technique is applied to improve the model robustness and forecasting accuracy. We compare our developed DLMs model with previously established models in terms of out-of-sample R square. Study of risk-return trade-off is also conducted based on our results.