Stocks classification using fuzzy clustering

Abstract

The main objective of this paper is to investigate the application of a pattern recognition technique as a supporting tool for stock investment decision taking by the public companies in the Brazilian Stock Market. The technique, known as fuzzy clustering means is applied to a set of indexes extracted from the quarterly financial statements relating to the 4 th quarter of 1994 through the 2nd quarter of 2002 belonging to oil/petrochemical and textile companies. The technique separates a group of companies into two sets. Having a set with higher potential returns than the other. And besides that, the set of stocks of the companies produced a higher potential yields and an average financial returns closer to the Bovespa index.