The objective of this thesis is to propose the use of a new tool for decision-making associated with the moment to buy or sell titles negotiated in stock exchanges. The proposed tool consists of applying a time series model to daily returns' logarithms and building a control chart with the model's residues (or forecasting errors), deciding to buy when the forecasting error is smaller than a lower limit (LL) and to sell when the forecasting error is greater than an upper limit (UL). A number of values for the above- mentioned limits (fractions of traditional values of limits for Shewhart and EWMA charts) were experimentally tested as a way to determine those which better meet the risk and return needs of investors with different degrees of risk aversion. The present work is an extension to the undergraduate degree thesis Control Chart Theory Applied to Decison-making in the Brazilian Stock Market (AMARAL, 2000) and the Master's dissertation Identification of Buying and Selling Moments of Stocks in Cash: an Alternative Procedure Inspired by Process Control Charts (AMARAL, 2004). The construction of charts for the residuals of time series models of Bovespa stocks' historical data yielded a tool which, in some aspects, was able to indicate moments for investment in stocks with better results than those obtained through certain stock funds. The indication of some adequate LL and UL to different types of investors (presented in chapter 7) was a concrete result which met the proposed objective. Complementary studies were made on the influence of seasonal effects, the result of the tool when applied to portfolios, the effect of the sample size, the degree of randomness of the results found and the behaviour of some stop-loss mechanisms. Great improvements have been made. However, given the degree of innovation of the work, it is recommended that advancements be made in certain aspects before including the tool amongst the techniques used by investors interested in variable income funds.