Font Size: a A A

Research On Stock Prediction Methods

Posted on:2007-11-01Degree:MasterType:Thesis
Country:ChinaCandidate:Y ZhuFull Text:PDF
GTID:2179360185963503Subject:Systems Engineering
Abstract/Summary:PDF Full Text Request
This thesis mainly discusses the problems in Chinese stock market prediction and the paper focuses on the following aspects:First, it studies the stock market prediction which is based on the fundamental plane system. It refers to the stock market as a complex system composed of many factors and a subsystem. Based on the investors' aim, needs, abilities and conditions offered by objective environment, people can creatively predict the future developing trend and the change in the stock market at a certain period of time.Second, the paper analyses the application of modern prediction method in stock market. It first points out the shortcomings of traditional analysis methods. Then it analyses the structure and standard flow of stock market prediction. In the end, the paper gives a detailed analysis and evaluation of the following prediction methods: the return prediction in current prediction method, the nero-network prediction and state space prediction and etc.Third, the combination model of state prediction method is studied in this paper. To determine Chinese stock tendency correctly, a state equation is given which is built up with stock tendency, steady self -regression AR model and nonlinear error covariant matrix. Based the state equation, a special prediction estimation method is devoted. Since all the value and tendency of stocks are known accurately when stock market is closed, the initial estimates of both the state and the error covariant matrix are determined. Thus, the new prediction estimation algorithm is simplified. Comparing with ordinary Kalman prediction method, the new ones can increase the accuracy of stock tendency prediction. Simulation to all stock data estimation and charger objections shows the new approach is more efficient than ordinary Kalman prediction method.Fourth, the paper also studies the application of state space prediction in the stock market analysis system. There are various kinds of softwares about the current stock markets. Each of them has its own advantages and disadvantages. Some softwares lacks of intelligence in analyzing stock market. Some list a number of exponents which are difficult for people to understand. Making use of the function of selective stocks and combining the fundamental plane system, technical exponents choosing shares with state space prediction, one can effectively improve their prediction without rearranging the index numbers and modifying parameters.
Keywords/Search Tags:Stock prediction, Negotiable securities analysis, Technical index choose of the stock, State space prediction, Return prediction
PDF Full Text Request
Related items