An optimizing utility for Portfolio Selection based on Optimal values computed using ANN, NSGA-II and Machine learning technique

Abstract

An optimizing utility for computing the values of derived economic factors of Portfolio selection is described in this paper. The significance of appropriate computing values of these factors has been felt because of the risk-constrained solution of portfolio selection. The classical Lagrangian multiplier method has been extended in the paper, using ANN and NSGA-II algorithm for computing weights used in the cost equations describing these economic factors. A mathematical formulation of the equations using portfolio selection parameters with computed values of weights is provided. A machine learning tool is given next for classifying values of coefficients of variations. Finally, a comparison of the ANN computations of weights with weights computed using NSGA-II is provided. This approach can be advantageous for the portfolio decision-making process.