Modeling the time-changing dependence in stock markets

The time-changing dependence in stock markets is investigated by assuming the multifractional process with random exponent (MPRE) as model for actual log price dynamics. By modeling its functional parameter S(t, ω) via the square root process (S.R.) a twofold aim is obtained. From one hand both the main financial and statistical properties shown by the estimated S(t) are captured by surrogates, on the other hand this capability reveals able to model the time-changing dependence shown by stocks or indexes. In particular, a new dynamical approach to interpreter market mechanisms is given. Empirical evidences are offered by analysing the behaviour of the daily closing prices of a very known index, the Industrial Average Dow Jones (DJIA), beginning on March,1990 and ending on February, 2005.