The data in financial markets are usually analyzed by the stochastic method for time series
analysis (Gardiner, 2003). The essential concept of this method was formulated in a study by
Einstein (1906) on the Brownian motion. In this article, Einstein studied the motion of a Brownian
particle in liquid by the thermodynamical method. Later, Langevin (1908) formulated this
phenomenon by the method of the Newtonian differential equation (Coffey et al., 2004).
It is common for economists to understand the Einstein theory of the Brownian motion in the
framework of Langevin.
To apply the Einstein’s method to financial data, we have to generalize Equation (1) to the
Generalized Langevin (GL) equation, which was introduced by Kubo (1966), and the white
noise (Equation 2) to colored noise (Hanggi and Jung, 1995). Corresponding to these
generalizations, we define a new Einstein relation. By the use of this relation, we can solve the
GL equation, calculate the autocorrelation function, and decide the probability density function
(pdf) of x( t). These functions are then fitted to the data, and we estimate the values of all the
parameters introduced in this model. |