Visualizing the log-periodic pattern before crashes

Abstract. We present a method for visualizing the pattern which we believe to be a precursor signature of nancial crashes (or ruptures). The log-periodicity of the pattern is investigated through the envelope function technique. Three periods of the Dow Jones Industrial Average (DJIA) are investigated: 1982{1987, 1992{1997 and 1993{1998. The presence of a rupture in the end of 1998 is outlined from data taken before the end of August 1998. PACS. 05.40.-a Fluctuation phenomena, random processes, noise, and Brownian motion { 47.53.+n Fractals 1 About a nancial crash occurence The application of statistical physics ideas to the forecast-ing of stock market behavior has been proposed long ago [1] and pursued in the pioneer work of physicists inter-ested by economy laws [2,3]. Econophysics [4] aims to ll the huge gap separating \empirical nance " and \econo-metric theories". Various subjects have been approached like the option pricing, stock market data analysis, mar