Steepest ascent hill climbing for portfolio selection

Abstract

The construction of a portfolio in the financial field is a problem faced by individuals and institutions worldwide. In this paper we present an approach to solve the portfolio selection problem with the Steepest Ascent Hill Climbing algorithm. There are many works reported in the literature that attempt to solve this problem using evolutionary methods. We analyze the quality of the solutions found by a simpler algorithm and show that its performance is similar to a Genetic Algorithm, a more complex method. Real world restrictions such as portfolio value and rounded lots are considered to give a realistic approach to the problem. © 2012 Springer-Verlag.