A Model Based On Support Vector Regression For Stock Market Predictions

Stock is considered as one of the most important financial tools of the financial market, whether the price of it could be predicted and how to predict it is always a controversial problem in the financial field. There are different kinds of models to predict the stock price fluctuation domestically and abroad. According to the different modeling theory, they could be divided into two categories:one is the structural econometric model based on statistical theories. Another one is the intelligent forecasting model based on neural network, genetic algorithm, and support vector machine. In this paper, we will apply theε-support vector regression model in the stock market prediction to give investors reasonable investment information.At first, we will introduce the support vector machine theory and describe the theory of Support Vector Regression, then by introducing theε-insensitive loss function, we will contribute a stock market prediction model based on theε-Support Vector Regression. In the part of the analysis, by introducing the data of Shanghai Composite Index, Sichuan Changhong and CMSB in a certain period and using LIBSVM software, we will make investigations and make error comparison between the predictions and actual values. At last, by comparingε-Support Vector Regression model and the traditional neural network method, we will show the feasibility and efficiency of the former one.How to predict the stock market effectively and accurately and enhance the profit is the problem which is concerned by experts and investors. The stock market prediction model we investigated in this paper is not difficult to understand, but also has good performance in predictions. So, what we have done in this paper has good effect on the prediction of the stock market of our country, it also has good effect on the right investment behaviors of the investors.