The Research Of Early Warning Model Of Chinese Stock Market Based On The Fractal Theory

This article bases on the fractal theory, studies early warning model of Chinese stock market.Comparing the date of shanghai and shenzhen with local hurst exponent,week average of local hurst exponent and month average of local hurst , put forward some early warning conditions based on china stock market.Using multifractal method combined with the cluster analysis,make a study on individual share. The mainly research works are included as following:(1)Comparing the date of shanghai and shenzhen with local hurst exponent,week average of local hurst exponent and month average of local hurst exponent, set up some early warning conditions based on china stock market.(2)Using the multifractal spectrum method to analyze the multifrctal characters of individual share's 5 min data,computing every shares'multifractal spectrum and their parameter, studying the changing character of multifractal spectrum and their parameter,we can find the regularity change of multifractal spectrum and their parameter in the time of share crash's approaching, beginning and termination.And then,we also ulteriorly find the linked relation between multifractal spectrum parameter and individual share yield.(3)Combining the parameter of multifractal spectrum,using cluster analysis method,we can classify these parameter as four typies and then make forecasting for them.We can find the rate of accuracy reaching 71.11% in the choosed interval.