Predicting Stock Price Using Support Vector Machines

The high-return characteristics of stock is attracting investors' attention.And they never stopped predicting stock price.Foreign stock markets have developed earlier,many economists have established different forecasting models through various methods.With the development of computer technology and mathematical theory,people began to use mathematical models to solve the problem of predicting stock price.Among them,the time series model is widely used and effective.With the development of statistical learning method theory,some machine learning methods such as support vector machines have begun to emerge,because of the use of the generalization error minimization principle and optimization methods,it attracts a lot of attention.This paper studies basis theory of the time series model and the support vector machines,and uses real stock data to fit model and analyze,and uses the mean square error index to compare the predictions of the two models.Time series analysis cannot interpret the linear part of the stock data and it is only effective in short-term prediction,while the support vector machine can handle the nonlinear problem well through the kernel function.This advantage can be reflected in the prediction result,the mean square error of the support vector machine's prediction results is much smaller than the time series model.It shows that the support vector machine performs well in stock price forecasting.