Identification of relevant factors and diversified portfolio selection in brazilian stock market

Starting from fundamental variables, this research aims to construct unobservable factors to explain the most part of the accounting data’s variability of listed companies on Bovespa. So, it was applied the multivariate statistical technique factor analysis by means principal components, in which eight factors were relevant. Based on the scores of these factors, the assets were ranked and annual portfolios with twelve actions were selected. This strategy had positive abnormal returns at 5% of significance level in the period 1999 to 2009, bringing evidence of inefficiencies in the Brazilian stock market. Thus, the techniques provide an option for understanding the stock market and the portfolios’ selection through the analysis of multiple variables