The price time series of the Italian government bonds (BTP) futures is
studied by means of scaling concepts originally developed for random
walks in statistical physics. The series of overnight price difference
s is mapped onto a one-dimensional random walk: the bond walk. The ana
lysis of the root mean square fluctuation function and of the auto-cor
relation function indicates the absence of both short- and long-range
correlations in the bond walk. A simple Monte Carlo simulation of a ra
ndom wall; with trinomial probability distribution is able to reproduc
e the main features of the bond walk. (C) 1998 Elsevier Science B.V. A
ll rights reserved.