Financial Sequence Analysis Based On Time Series Model

It was considered that the price volatility of stock market was a kind of Brown motion and it followed the law of random walk in early capital market, so Efficient Market Hypothesis(EMH) was put forward. EMH is the foundation of a series of modern classic theories in finance, and it has been applied widely on the research in the capital market. However, with the development of non-linear theories, People have begun to doubt the basis of the theories. Most of the traditional capital market theories are based on linear theories, while the variables in actual capital market are non-linear, so the traditional capital market theories are unable to explain, non-linear methods become the leading actor. Capital market is chaos. There are many models of financial series. One of the main model is VAR(p) model which is used to process multiple variables of time series. It is easy to analyses and forecast. Support vector machines an effective tool for identification. Support vector classification is an effective tool to classify parameters of the model, thus it show advantages in many areas and have successful applications.This paper includes four chapters.Firstly, it is an overview of the non-linear theories, including the background and research status of chaos,fractal theory,vector time series model,support vector machines. Secondly, it was put forward the chaotic of the financial markets. We calculate the fractal dimension of the stock market of shanghai and Shenzhen as two examples. The fractal dimension focused on 2-3, so we test and verify the existence of chaos. Then we suggest the construction of VAR model. Thirdly, we establish the VAR model.The processing of the sample data mainly use Matlab and Eviews. The financial series are forecasted in the fourth part and we use SVC for the identification of the model. The results are obvious. The last section is a summary of this paper.This paper has the following innovations:1) According to the characteristics of the financial markets, we calculate the fractal dimension for series of the day closing price of Shanghai and Shenzhen by using Matlab program, thus the existence of chaos are verified.2) We established VAR(p) model,estimated the parameters of the model and predicted the series.3) We realized the classification by SVC, the rustles of the identification are obvious and zero error. This discovery will help us to understand the financial market of China with new point of view, and guide us to grasp the rules of the market better.