Intelligent System for Portfolio Selection

The aim of this paper was to develop an intelligent system for portfolio selection that assists the investor in selecting assets for the composition of an optimal portfolio of investments. It was built a variation of the Markowitz Model, where the forecast price is reported by a predictor, using the Support Vector Machines (SVM) technique. The SVMs obtained an average prediction error of 7.13% and a standard deviation of 2.88%, which shows that most of SVMs performed good predictions about the data set.