Refining asset allocation under the resampled mean-variance optimisation approach: Evidence from the Johannesburg Securities Exchange

Abstract

The introduction of mean-variance optimisation by [1] has resulted in a profound interest in asset allocation amongst researchers and practitioners alike. It is argued to be the most important decision within the investment decision-making process. It is therefore no surprise that the process of asset allocation optimisation has evolved to a great extent over the years, and researchers continue to examine ways in which it can be further improved. During the last two decades, a number of researchers have observed that the resampled mean-variance optimisation approach offers superior results. In this article we examine the effect of using three parametric (the normal, student-t and mixture) and two non-parametric (bootstrap and weighted bootstrap) sampling techniques in an attempt to further refine the resampled optimisation approach. To align our results with one of the most recent studies, we use the Johannesburg Securities Exchange (JSE) as our subject stock market. Our results suggest that a student-t sampling technique is superior for a ‘risk-neutral’ investor while a normal mixture sampling technique performs best from a risk-adjusted return point of view.