ENBIS: European Network for Business and Industrial Statistics

Method for VaR-based Portfolio Selection

22 September 2008, 11:43 – 11:47

Abstract

Submitted by

Peggy Ng

Authors

Xu, Chunhui and Ng, Peggy

Abstract

Value-at-Risk (VaR) is taken as a measure of risk in portfolio selection. Since VaR is generally a non-convex and non-smooth function, conventional optimization methods fail to solve portfolio selection models that incorporate VaR. In this paper, we propose a soft optimization method for solving VaR-based portfolio selection models. In theory, the soft method does not necessarily produce an optimal portfolio, but identifies a good enough portfolio with a high probability. In return for this compromise, the soft method is simpler to use than linear programming algorithms.
An investment experiment is performed using data from the New York stock market. Portfolios suggested by the soft method are compared to two commonly used investment strategies. The results show that the portfolio selected by the soft method may produce a higher return at lower risk than the two investment strategies, which demonstrates the effectiveness of the proposed method. The results also suggest that VaR could be applied as a measure of risk in portfolio selection from a computational point of view.

Return to programme