The trend analysis of China´s stock market based on fractal method and BP neural network model

Traditional mainstream financial theories usually focus on the assumption of rational man. Many scholars often suppose that stock market is the efficient market and security price fully reflects information gained by investors from stock market. But with financial anomalies emerging, many new subjects related to physics, psychology and computer science etc., are rising and forming behavioral finance, financial physics and network finance. This paper using fractal analysis method studies fractal properties of China´s stock price fluctuation and arrives at a conclusion that market trend would be slightly overturned in the case of Hurst index value lower than a critical point. Then BP neural network is used for fitting the trend of China´s stock price fluctuation. We construct a three-layer neural network that is applied to weight adjustment between neurons located in different layers. Experiment results show that the prediction curve of network outputs has a better fitting with the real curve of original sample data, so BP neural network in this paper has the preferable predication performance.