Hybrid search for cardinality constrained portfolio optimization

In this paper, we describe how a genetic algorithm approach added to a simulated annealing (SA) process offers a better alternative to find the mean variance frontier in the portfolio selection process. The nonlinear mixed integer quadratic programming model is considerably more difficult to solve than the original model; but some computational experiments have shown that hybrid heuristics offer a good alternative for these types of problems. Categories and Subject Descriptors I. Computing Methodologies I.6. Simulation and modeling. I.6.5. Model Development General Terms Management. Keywords Portfolio selection, Markowitz model, Mixed integer programming.

• Type: Conference
• Year: 2006
• Where: GECCO
• Authors: Miguel A. Gomez, Carmen X. Flores, Maria A. Osorio
• Keywords: GECCO 2006, Integer Quadratic Programming, Mixed Integer, Optimization, Portfolio Selection