This paper is about one of the most argued subjects in the financial theory: the forecast of future returns. We work with the model of multifactor of Fama and French, and the regression presented for Grinblatt and Moskowitz (2002), that work with the size of the company and the book-to-value. We present here also an ample analysis of the monthly autocorrelations between the most negotiated shares in Brazil from 1995. The joined results suggest that the analysis of multifactors of Fama and French shows sufficiently useful and with variables like size and book-to value being significant. In relation to the autocorrelations, in the majority of the cases if they had not shown individually significant, what it suggests efficiency in the weak form? The results suggest that the model presented by Grinblatt and Moskowitz (2002) won´t show good results, in the first moment, strategy of abnormal returns only with time series data.