Abstract
This paper applies two computational intelligence techniques, namely particle swarm optimization and artificial immune systems, to constrained portfolio optimization. The portfolio selection model considered in this paper is based on the classical Markowitz mean-variance theory enhanced with floor and ceiling constraints. Several experiments are conducted using the stocks listed on the Karachi Stock Exchange 30 Index (KSE30). The performances of both computational intelligence techniques are compared on two criteria: (a) maximization expected return and (b) maximization of return-to-variance ratio. The results are also compared with the ones obtained through Microsoft Excel Solver.