The Application Of Time Series Analysis Methods In China's Stock Market Forecasting

In recent years, the forecast based on the analysis of time series, is widely used in various fields. China's stock market is a highly complex nonlinear dynamic system. In China, people usually use time series model to predict the stock price, but later scientists propose neural network model by studying human brain. Due to the self-adaptive study, highly robustness and fault tolerance of neural network, and it's fully approximation to complicated nonlinear relation, more and more attention is being paid to the neural network model and the model is accepted by people. In addition, the neural network model achieves remarkable results in stock forecast.The time series of stock market with the following two properties:In the first place, it seems random but incomplete answer; in the next place, it is easy to get. Therefore, many studies and scholars as well as the stock traders are hope of finding some rules to predict the stock price or stock revenue position exactly. Time series analysis methods is the latest quantitative forecasting methods, it is particularly applicable to the time series, because of the economy involve a lot of factors, and more complex, it is difficult to predict numerical rational model. The text mainly discuss stock-price forecasting model. Time series predict methods reflect the long-term trends of stock price and the price of the short-term technical adjustments of stock price can be analysis in neural network model for its nonlinear.This paper introduces the concepts of time series analysis, then presents three time series econometric model, and finally introduces another time series forecasting model - neural network model, for neural network model can fit any nonlinear curve. The idea is predictability in a given sample through the data for a given machine training and establish a relationship between output and input functions. This paper will chose individual equities of China's stock market data - YaTai group and KangMei pharmaceutical industry as the research object, and according to the short-term investment idea of investors, to predict the future stock price movements. This paper adopts the methods realized by Matlab software for testing the Chinese stock market data, through the toolbox and a simple programming environment of the software. Combination of mathematics knowledge that we have learned, we choose the linear time series model and BP neural network model to simulate and predict the stock index data again. If we compare the forecast result of linear model with the forecast result of BP neural network model through error analysis. We will know the advantages and disadvantages of the linear model and BP neural network model by comparison, and provide valuable suggestions for the future work.