Title: Beating the market: Can evolutionary-based portfolio optimisation outperform the Talmudic diversification strategy?

Authors: Safwan Mohd Nor; Sardar M.N. Islam

Addresses: University of Malaysia Terengganu, 21030 Kuala Nerus, Terengganu, Malaysia; Victoria Institute of Strategic Economic Studies, Victoria University, Vic 3000, Australia ' Victoria Institute of Strategic Economic Studies, Victoria University, Vic 3000, Australia

Abstract: It is argued that with a small number of stocks (N) in a portfolio (which suits individuals rather than institutional investors), naive Talmudic diversification rule (1/N) offers a superior trading outcome against mathematically optimal portfolios due to its robustness against estimation error. As this puzzle has not been resolved, we explore it using an alternative portfolio choice problem that seeks to outperform the benchmark market index - FTSE Bursa Malaysia KLCI. This study makes a significant contribution by using an industry-common objective function and also incorporating floor/ceiling constraints and the effect of delisting. Using evolutionary algorithm, we construct optimal portfolios with varying Ns in-sample for out-of-sample analysis. We find that 1/N is superior with smaller Ns, although optimised portfolio dominates as N increases. However, with both diversification policies underperform the market and produce very low Sharpe ratios, their efficacies for practical applications are highly suspect.

Keywords: portfolio optimisation; Talmudic diversification; realistic trading constraints; individual investors; institutional investors; evolutionary algorithms; portfolio selection; floor-ceiling constraints; delisting.

DOI: 10.1504/IJMEF.2016.074584

International Journal of Monetary Economics and Finance, 2016 Vol.9 No.1, pp.90 - 99

Received: 14 Jul 2015
Accepted: 19 Aug 2015
Published online: 06 Feb 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article