This dissertation approaches a very recent and controversial issue named Behavioral Finance. So, this work begins presenting some of the concepts and studies carried out in the area, where some criticism of the efficient market hypothesis and the random walk idea is made. This criticism drives to another three concepts: autocorrelation of asset return, mean reversion and predictability of asset return, which, indeed, are the central issues of this work. To explore these three concepts the Box&Jenkins model will be applied on daily return time series of the most 50 liquid stocks listed in the São Paulo Stock Exchange (BOVESPA), between 01/011994 thru 12/31/2005. At the end, the study concludes that exist autocorrelation evidences among daily returns, that there is a possible indication of mean reversion, and that the forecast model based on past results has just a regular performance.