The financial crisis and the market uncertainty of the last years have pointed out the shortcomings of traditional portfolio theory to adequately manage the different sources of risk of the investment process. This paper addresses the issue by developing an alternative portfolio design, that integrates risk parity into the cardinality constrained portfolio optimization model. The resulting mixed integer programming problem is handled by an improved multi-objective particle swarm optimization algorithm. Three hybrid approaches, based on a repair mechanism and different versions of the constrained-domination principle, are proposed to handle constraints. The efficiency of the algorithm and the effectiveness of the solution approaches are assessed through a set of numerical examples. Moreover, the benefits of adopting the proposed strategy instead of the cardinality constrained mean-variance approach are validated in an out-of-sample experiment.
