Support Vector Machine And The Applicaiton On The Prediction Of Stock Price

Stock market is a complicated non-linear dynamic system. It is very difficult to open out it's inherent rules using traditionary timing prediction technique.First, I will develop a new method, which is upon the support vector machine(SVM), to solve the problem in the aspect of forecasting the stock price. And then I will optimize the arithmetic of support vector machine, using the effective constrain method for quadratic programming, and based on the theory of chunking arithmetic in support vector machine. As the result, the optimized arithmetic——single-dimension chunking arithmetic(SDC), is deduced. The new method will be applied on solving all the SVM problem in this paper, and the result is very good. Subsequently, I will establish the Simple Forecast Model(SFM) to solve the stock price forecast problem. Then I seriate the decision function in the progress of SVM, accordingly the Simple Forecast arithmetic (SFA), which is used to solved the SFM,will be developed.In this paper, I make lots of experiments with SFA, and point out that what should be given attention when using SFA to solve stock price forecasting problem. In the end of this paper, I will deduce a more complicated model based on the SFM, named Bounced Forecast Model (BFM). BFM overcomes the weakness of SFM, but I also point out that BFM has a bigger problem. Accordingly, SFM is approved to be a good model to solve stock price forecasting problem.The data that is used in this paper is the real data from some typical stocks in our country, and all the models is established for analyzing these data.