An Integrated Model for Stock Price Prediction Based on Svm and Arma

Changes in stock price will be influenced by many aspects of factors. When we are predicting stock price, it is difficult to build a determined mathematical model between stock prices and these complex factors. This paper first utilizes ε − SVM (ε − support vector machine) to build a stock price prediction model. By fitting the prediction error sequence, we find the law factors, which the prediction model could not include or failed to give sufficient explanation, have a lasting influence on the stock price according to the ARMA theory of the time series analysis. Through an analysis we can predict the changes in the next step of this unexplained impact. Thus we revise the SVM model in the previous step. And the final results are obtained. The empirical analysis proves: ε − SVM model has a high prediction accuracy to predict the stock price. Moreover, after the ARMA error correction model is used, the prediction accuracy is further improved.