Portfolio Optimization Without Optimization

Abstract

In this paper we introduce a simple, yet flexible approach to construct (enhanced) index tracking portfolios. PADME (Portfolio Allocation via Density MatchEs) uses the density function of returns as a robust means of capturing investor preferences, and identifies suitable portfolios through a measure of closeness. Recognizing model uncertainty, PADME avoids optimization and instead relies upon the behavioral concept of revealed preferences by offering the investor a palette of options from which to choose. In a case study we consider an EIT manager with an SP500 benchmark. We illustrate PADME's flexibility by imposing several types of Bayesian beliefs simultaneously. Using far fewer assets than the SP500, we are able to generate numerous portfolios whose out-of-sample risk and return were superior to the benchmark during the turbulent COVID Crisis of 2020.

Keywords: index tracking, portfolio resilience, evolutionary heuristic, revealed preferences, full scale optimization

JEL Classification: G11, G02