The fluctuations of financial markets appear, in general, to be unpredictable in advance, and caused by the stochastic, chance arrival of 'news' and other shocks. There is, however, a possibility that the market may be subject to a non-linear dynamic process, which can appear to mimic such stochastic processes. Such non-linear dynamics can, however, on occasions generate major breaks, disturbances, which is one reason why such systems have been given the somewhat lurid title of 'chaotic'.
One difficulty of testing for the existence of such 'chaotic' systems is that the available tests require a vast amount of data for tests to have sufficient statistical power, requiring upwards of 10,000 data points. In this preliminary test for chaos in the foreign exchange market, Vassilicos uses a very large data set of consecutive price entries for the Dm/$ exchange rate, with well in excess of the required number of data points.
Vassilicos applies one of the latest available tests for the identification of non-linear dynamics to this data set, and concludes that "the variations of the last two decimal digits of the Dollar/D.Mark rate (i.e. the high frequency fluctuations of that exchange rate) are not caused by low dimensional chaos". He notes, however, that this result needs further corroboration with other tests, (eg. one based on Lyapunov exponents), and similar tests on other forex (and other financial) series
Download is not available
