摘要:This study develops a method to detect trend reversals followed by significant drops in Brazilian Stock Market using wavelets. Applying the concept of the log-periodic power-law, wh ose oscillations present reduction in amplitude and period as the critical moment approaches (where there is a higher probability of market drop), it used the Continuous Wavelet Transform to detect the increasing oscillation frequency in the stock price, time series and generate sell signals. An algorithm w as developed to test different kinds of wavelets and parameters to calculate the Wavelet Index, evaluating wh ose combination of parameters brings the best results and comparing these results with the existing Technical Analysis tools. The results show that the proposed method to cal cu late the Wavelet Index detects successfully the significant drops (over 10%) in the Brazilian Stock Market. Considering drops over 15%, there were losses due to early sales of 45% (average) in the search set and 43 . 9% in the test set, without false negatives, using mainly the Meyer wavelet. Its performance was also better than existing Technical Analysis tools, like MACD and RSI.