Empirical Evidence on the Predictability of Stock Market Cycles: the Behaviour of the Dow Jones Index Industrial Average in the Stock Market Crises of 1929, 1987 and 2007.

Based on a deterministic hypothesis, this paper aims to verify the regularity of the stock market cycles and, if this regularity is found, the ability to predict major stock market crises. Harmonic analysis, or Fourier series, is applied in order to, decomposing into sinusoids curves, find the constant periodicities hidden under the series of observed data. Starting from the industrial stock market data in the U.S., considering three periods of similar length of 165 months: 1919:01 to 1932:09, 1977:01 to 1999:09 and 1997:03 to 2010:11, I stand in the moment of maximum growth of the Dow Jones Industrial Average and I check if the most significant hidden periodicities allowed to predict the sharp drop in the index that was coming and the subsequent development. The evidence is inconclusive. A small number of theoretical cycles reasonably explain the stock market evolution. In terms of predictive power, in two cases there is this ability, while not in another. The conclusion reached indicates that, due to the regularity in the data, the application of the a deterministic hypothesis is reasonable. However, it is necessary to perform a deeper analysis of the data to be able to describe and predict major stock market cycles, including crises or large declines in stock market prices.