Large earthquakes, rupture in complex materials, stock market crashes: all can be viewed as catastrophes - the sudden transition from a quit state to a crisis. Would it be possible to forecast these events? A unified framework for the modeling and forecasting of catastrophes was proposed by D. Sornette, based on the concept of log-periodicity. In this article we discuss the potential for predictability of this theory and we illustrate it in problems related to stock market crashes. We study this method of forecasting applied to the the stock market index of the BOVESPA São Paulo Stock Exchange, IBOVESPA. We seek evidence of log-periodic behavior, comparing a period without crashes to the period before the crash of 14 January 1999. The efficiency and the relative simplicity of the method serve as incentive to undergraduate students eager to see the theory being put to practice.
scale laws; log-periodic; catastrophe
