The Zipf analysis of n-words in random sequences and financial data series
like the stock prices of a company has been performed. The bias as well as
the resulting staircase structure of the Zipf plots are taken into account
in the subsequent analysis, It is found that correlations for the sign of t
he fluctuations as well as for the amplitude of the fluctuations can be fou
nd in financial time series. The relevance of the n-Zipf analysis to financ
ial sequences is underlined to be only weakly predictive for a "binary tran
sformation level", but could be more interesting for "higher translation le
vels", (C) 1999 Elsevier Science B.V. All rights reserved.