Essays on characteristic-based portfolio optimization

Abstract

In my dissertation, I develop new portfolio optimization methodologies designed to enhance empirical performance in practical applications. Instead of relying on asset characteristics as intermediary variables to estimate the distribution of asset returns—an approach that introduces estimation errors—this work explores methods that directly incorporate asset characteristics as input variables in the optimization program. By bypassing the estimation step, these approaches aim to improve robustness and efficiency in portfolio construction. This dissertation contributes to both active and passive portfolio management.