Long-memory behavior analysis of China stock market based on Hurst exponent

摘要

Previous mathematical models shown by empirical studies, such as Poisson model and Markov Process based on the hypothesis of short-dependence and independent arriving features of stocks data, failed to describe the stock market's behavioral characteristics with high accuracy. In this paper, we first offered thought for fractal theory and introduced an important index in this theory, Hurst exponent (H), to better depict stock movement in a more objective way. Then we selected 5 China's representative stocks in different years and applied R/S analysis on them to calculate the value of Hurst exponent. The result, Hurst exponent of the daily price of stocks is larger than 0.5, which implicates the rejection of previous models, proved that the stock market follows the Fractal Market Hypothesis (FMH), and stock tendency has the long-range dependent and fractal features. Thus, Hurst exponent provides a better measure with a new perspective to analyze the overall stock market and also gives useful guidance for investors.