Analysis And Forecast Of China's Stock Market Based On Fractal And Chaos Theories

This article bases on the non-linear theories such as fractal and chaos, studies in the rules existed in of our stock market price process from the non-linear theory visual angle, which provides some reference methods and the apocalypse for investors and financial supervisor. The mainly research works are included below:(1) Uses the newly V/S method, proposed by Cajueiro (2005) and so on to estimate Hurst exponent, and the Hurst's classics R/S analysis to study the long memory of CSI300 index, which delegates the fractal characteristic of our country stock market at present. Based on the V/S analysis, the return series of CSI300 index has a Hurst exponent of 0.55, the Hurst exponent of its volatility series is 0.76, which are bigger than 0.5, and the non-circulation cycle of return series is about 199 days. The demonstration results indicate that our country stock market has obvious long memory, and it is a fractal market.(2) Uses a new risk measure method for fractal market proposed by Hong-quan Li, to inspect the risk of CSI300 index. And find that the standardized risk vectors of the entire sample series, the series of January, 2002 to December, 2004, and the series of January, 2005 to October, 2007 are (1.1, 1.088, 1.524), (1.122, 1.356, 1.258) and (1.058, 0.921, 1.419), the corresponding risk assessment index is 1.237, 1.245 and 1.133 as well. The analysis result indicates that our country stock market unceasingly consummates, the market risk reduces day by day, but at present still had a higher risk.(3) Modifies the converted fractal model, used in price forecast by Yuhua Fu, in the demonstration of the predictability of CSI300 index, and it gets the better forecast effect during about 5 days. We also joins a positive feedback factor into the input level of BP Neuron Network to form the new Positive Feedback-BP Neuron Network, because of the existence of positive feedback trade in financial markets, which is certified by behavior finance, and it has quite ideal forecast effect in 10 days'forecast.(4) Discusses the problems which our country stock market still exists at present, and proposes some enlightenment to financial supervisor, based on the demonstration.