Multi-criteria Portfolio Optimization Using Support Vector Machines and Genetic Algorithms

Abstract

Given that not all the assets available in the market are appropriate for a given investor, it is desirable to stratify these assets into different classes on the basis of some predefined characteristics. Furthermore, using investor preferences, one needs to select some good quality assets from a given class to build an optimal portfolio. The focus of this chapter is to present a hybrid approach to portfolio selection using investor preferences in terms of selection of assets from a particular class that suits the given investor-type. The support vector machine (SVM) with radial basis function kernel is used to classify the assets into three classes. The optimal portfolio selection is achieved using a model that is based on four financial criteria: short term return, long term return, risk, and liquidity. A real coded genetic algorithm (RCGA) is designed to solve the portfolio selection model.