Analysis Of The Stock Market By The Suppot Vector Machine And The GARCH Model

With the development of the economy and the rise of people's attention to the stock market,the stock market has become a “barometer”of the economic field and indicated the trend of a country's economy.The price fluctuation of the stock market are,however,extraordinarily unpredictable due to its features of complication,nonlinearity and uncertainties,so the traditional structural model of measuring and the time series model can not meet the needs of prediction for stock data.The support vector machine is a new machine learning technique.It has established on the bases of statistical study theory,optimization theory and minimzed structural risk principle,which can better handle the issue of small sample learning meanwhile avoiding the defects of “dimension disaster”,the slowness of web convergence and easily causing local minimum.At present this tech has successfully applied in several fields,used mainly in the classification of data and the regression problem.With the continued deepening of financial series study,people gradually find the relative accuracy of the autoregression conditionally heteroscedastic model in depicting the market fluctuation with certain predictive ability.In order to accurately recognizing and measuring financial market risk,it's necessary to introduce VaR into our stock market.All of these have a significant reference value for analyzing the implication of stock-market economy and guarding against financial risks.This article firstly introduces the concerned theories of the support vector machine and then makes the explaination on the time series model and the venture worth.Through selecting CSI 300 Index and the prices of individual stocks by treating them as study objects,this article makes a prediction on the realistic conditions of building the conditional heteroscedastic model and using the VaR value of CARCH model to depict stock market risks.By constantly optimization,the parameters of the support vector regression model have eventually established,also applied a integration of the support vector regression calculation and the condtional heteroscedastic model to make a prediction on CSI 300 Index and a spectulation on the trend of its future.Finally,through the comparation between the predicted value and realistic value,we has achieved both prediction accuracy and expected results from several models.The upshot of this article shows: The prediction accuracy of the support vector machine is superior to other models,and its prediction results on share data are quite satisfiable,thus we can see that support vector regression calculation has certain of effectiveness on the stock market predition.