Simple rules or optimization for a dollar-neutral investor?

Abstract

In this thesis, we study portfolio construction and asset allocation for a long/short investor. We construct equally weighted portfolios based on known firm characteristics and compare these to mean-variance optimization models in two different datasets. We find that high turnover and estimation error diminish the effects of optimization after transaction costs. Simple median-based 1/N strategies are not necessarily optimal, but all strategies manage to outperform mean-variance models in the sample consisting of a larger number of assets. Further, the medianbased 1/N strategies we consider could be used as potential benchmarks for active characteristic-based strategies.