| Article ID: | iaor19991836 |
| Country: | Belgium |
| Volume: | 37 |
| Issue: | 3 |
| Start Page Number: | 23 |
| End Page Number: | 40 |
| Publication Date: | Jan 1997 |
| Journal: | Belgian Journal of Operations Research, Statistics and Computer Science |
| Authors: | Hamza Faris, Janssen Jacques, Ruelas Gabriel |
| Keywords: | optimization, programming: quadratic |
In this paper, an application of large scale portfolio optimization problems based on mean-semivariances criterion is presented. It is illustrated by a numerical computation based on real data set of returns, which represents the complete Belgian Stock Market. We discuss the comparison of the optimal portfolios given by different models obtained by a suitable selection of the risk function parameters, when pattern of investment returns is not symmetrical. Ex-post performance differences between portfolios are also examined using different time horizons. All quadratic programming computations and efficient frontiers are calculated by using the library OSL installed on the Risc System/6000 workstation.