Portfolio optimization in the Kenyan stock market: a comparison between mean-variance optimization and threshold accepting

Abstract

The Mean-Variance Optimization (MVO) model has been used in asset allocation problems since the inception of Modern Portfolio Theory in 1952. Several improvements and alternatives to MVO have been suggested and used since then. These include adding constraints to the traditional MVO model, using alternative risk measures and use of non risk-reward models. This study seeks to compare this risk-reward model against the Threshold Accepting model, which is a general optimization model, in portfolio selection in the Kenyan stock market to establish optimal stock portfolios to be held by investors in The Nairobi Securities Exchange (NSE). A comparison is done between the two models by measuring their performance using the following performance ratios: Sharpe Ratio, Sortino Ratio and Information Ratio using 29 stocks in the NSE from 1998 - 2016. Using portfolio performance ratios, it is concluded that the Threshold Accepting (TA) model outperforms the Mean-Variance Optimization model but the latter is observed as a more consistent model. The TA model has portfolios with generally more superior returns relative to the risk taken for the full period; however, this is not consistent over varying time estimates. This observation implies that attention should be given to the TA model rather than the classical MVO approach with the aim of improving optimal portfolio selection.