A Study On The Predictability Of China's Stock Exchange Based On Time Series Methods

Time series from stock exchange has the following characteristics. First, it appears to be a random walk but may be not totally random. Second, it is easily and almost costlessly available. Therefore, lots of researchers and stock traders hope to find some rules in it in order to make accurate prediction of stock price or return rate. Time series method is a newly-developed quantitative method for prediction and yields satisfactory results in the analysis of economic time series in which the involved factors are too many and the relationships between them are too complicated, leading to the application of theory-based quantitative predicting methods unworkable.Like stock exchanges in developed countries, China's security exchange optimizes productive resources but unlike those in developed countries, it is highly volatile. In this essay, firstly the author analyzes the predictability of time series from China's stock exchange using three kinds of methods: ARMA model, neural network model and non-parametric estimation and gives evaluation on their performances while at the same time puts forward some conclusions deserving attention from both stock exchange supervising department and stock traders. Secondly, the author examines the assumptions closely on which the above-said methods base and gives a detailed discussion on them, especially using GARCH model to test quantitatively the stability of China's stock exchange, afterwards drawing the conclusion that it is hard to make accurate prediction of price or return rate of China's stocks for none of the assumptions fully holds ground. Thirdly, taking account of the difference between Chinese stock traders as a whole and that of developed countries, the author gives a thorough analysis on the complexity and volatility of its (traders') reaction to information and points out that the intrinsic heterogeneous and volatile reaction to information is an important reason for the almost unpredictability of the price or return rate in China's stock exchange. Given that individual-dominated situation in China's stock exchange will last for a long time, it is of great importance to integrate behavioral financial theory into the quantitative research on the time series from China's stock exchange. Finally, the author points out the drawbacks of phenomenon-based predicting and presents an outlook on the direction of the analysis of time series from China's stock exchange in view of the difference between theory-based predicting and phenomenon-based predicting.