A chance-constrained approach to stock selection in Hong Kong

Abstract

To deal with the uncertainty of models a stochastic programming technique, chance-constrained programming, developed in 1959 by Charnes and Cooper, is employed to modify the classical model of Markowitz so that it is more realistic. Chance-constrained programming does not require that a constraint should always hold but only be satisfied with a given probability. In this paper the Markowitz model is first modified with the chance-constrained programming technique by determining the randomness that exists in the average rates of return of stocks. This modification requires the assumption of the normal distribution of return rates to find an estimator of the variance of the portfolio. Then the modified model is run for four satisfaction levels: 99, 95, 90 and 85%, together with the classical one, at different required rates of return.

Notes

Tel: + 1 [phone]; Fax: + 1 [phone].