Empirical autocorrelation functions of returns
of stochastic price processes
show phenomena of correlation on small intervals
of time, which decay to zero after a short time.
The paper deals with the concept of weakly correlated random processes
to describe a mathematical model which
takes into account this behaviour of statistical data.
Weakly correlated functions have been applied to model
numerous problems of physics and engineering. The main idea is,
that the values of the functions at two points are uncorrelated if
the distance between the points exceeds a certain quantity
epsilon>0. In contrast to the white noise model, for distances smaller than epsilon a correlation
between the values is permitted.
A property of the introduced price model is, that the corresponding stochastic
processes possess absolutely continuous sample paths.
Therefore the used concept implies,
that the so called condition of ''No Free Lunch with Vanishing Risk'' is
not fulfilled.