Research And Computer Implementation Of Stock Price Forecasting Methods

Stock market is full of risk. The author attempts to forecast stock price and guide the investment effectively by investigating the potential low of stock price, which is a timely waved series. The stock price is sensitive to a lot of factors including those of its owner and its surrounding. So compared with other forecast systems, the stock price forecast system needs more consideration. The thesis focuses on this research with two sides-qualitative analysis and quantitative analysis, mainly on quantitative analysis which attempts to construct suitable models to describe the behavior of stock price. And finally merges these two kinds of analysis and puts forward a systematic method of forecasting the stock price.Qualitative analysis is based on traditional analysis, which is composed of basic analysis and technology analysis. Basic analysis deals with the problems of which stock should be bought and when to buy, by analyzing the Macroeconomics, trade and company condition. Then using technology analysis tries to find the development trends, pressure and support of stock price.Quantitative analysis is the second module in the stock price forecast system (The first module is solving the problem of what kind of stock should be chosen.). The stock price forecast model has been built by utilizing mathematic methods and CAD. In this model the error is also be controlled within the specific limits. In the paper the author analyzes stock price from linear model to non-linear model and builds three models: time series model, neural network model and gray neural network model. They improve precision one by one. Case citing in the paper also proves the model's validity.In this paper, the author summarizes a lot of useful laws of the model-building and hopes them can be helpful to other investigators. At the end of paper the author points out the direction of the system's development.