Non-Gaussian optimization model for systematic portfolio allocation: How to take advantage of market turbulence?

Issue title: Computational Techniques in Finance

Article type: Research Article

Authors: Ndiaye, Papa Momar

Affiliations: 40 River Road, Suite 14B, New York, NY 10044, USA. E-mails: pmomar@gmail.com, pape@raisepartner.com, URL: http://www.raisepartner.com

Abstract: In this paper, we show how to build a systematic quantitative portfolio allocation strategy using non-Gaussian risk metrics and market turbulence detection. We propose a 3-step approach to build a reference second-order portfolio that will be optimally rebalanced according to increasing market skewness. The reference second-order portfolio uses a robust estimate of a covariance based on a semi-definite matrix calibration and incorporates risk aversion via a moving performance target depending on the volatility of the VIX index. As such, it is a dynamic strategy balancing robustness and reactivity to market volatility changes. A convex relaxation scheme allows to re-formulate the mixed second-order/third-order optimization problem as a tractable semi-definite program.

Keywords: Markowitz optimization, semi-definite least square, positive asymmetry, convex relaxation

DOI: 10.3233/RDA-2011-0048

Journal: Risk and Decision Analysis, vol. 2, no. 4, pp. 237-244, 2011