Crises or crashes in financial markets have been studied since the well known event of 1929 and several mathematical models have been proposed in the literature in order to be employed in the process of forecasting abrupt changes in the behavior of investors. Methods involving econometric models are quite common in works dealing with quantitative tools for analysis and forecasting of financial scenarios. The main idea in this work is to present one of these traditional methods, namely that using a log-periodic model and compare its performance with a new proposed method that uses Wavelet Transforms to detect changes in the price trends. An evaluation was carried out using historical pre and post crash data of the 1929 event, as well as more recent data from Dow Jones Average (USA), Hang Seng Index (Hong Kong) and Ibovespa (Brazil). These data were used to test the adequacy of the early warning of occurrence of crises in financial markets, as provided by the two methods.