It has been observed that at the large time scales the distribution of stock market returns is convergent from Boltzmann distribution to Gaussian asymptotic one, To explain this universal phenomenon, the authors propose a new and simple dynamic model to describe this convergence by the time parameter in association with the introducing the concept of relaxation time for financial markets. The analysis of stock market data packages in different time intervals showed that the model fits well the financial market data. The meaning of so-called relaxation time has been qualitatively made clear, as a measure to estimate the stability of the market.