Mean-ETL Portfolio Selection under Maximum Weight and Turnover Constraints Based on Fundamental Security Factors

Abstract

In this article, we model stock returns using fundamental data and minimizing average value at risk (AVaR) and multiperiod portfolio selection with weight and turnover constraints. Equity returns are decomposed into returns explained by fundamental and nonfundamental factors. While the former are found to be independent, the latter are found to be highly dependent among various stocks. Then, we construct models to forecast returns using several ARMA–GARCH models with different innovation distributions and simulate scenarios of future returns. Based on these scenarios, we examine various approaches of portfolio optimization. By comparing actual portfolios based on real data, we find that 1) the ARMA–GARCH model with classical tempered stable distribution provides a superior prediction of equity prices than the normal and Student’s t-distribution and 2) AVaR provides a better risk measure than variance. We also see how portfolio performance changes under weight and turnover constraints and suggest that it is effective to reduce the stock universe and trade large-capitalization securities.