Price Prediction Based On Support Vector Regression Model

Based on machine learning,this paper uses support vector machine to study the regression forecast of stock index and national bond index future closing price.The focus is mainly on asset price forecasting based on support vector regression model.Firstly,in order to verify the advantages of the SVM model in solving small sample models,-SVR model,the radial basis kernel function was used to analyze the different sliding window lengths of the sample data.The results show that the longer the sample sliding window,the greater the average absolute error,but the accuracy of the prediction direction does not change significantly.Secondly,based on the selection of the length of the sliding window,the data processing and model structure selection involved in the model are analyzed in turn.Then,the parameters of the selected model are optimized using the improved grid method,and the final results is more accurate prediction model.Finally,seven technical indicators were selected to construct a new index system.Compared with the index system constructed using basic factors,it was found that the regression error rate of the model established using the technical indicators was less than the regression error rate of using the basic indicators.For different index systems,the selection of data processing methods and model structures is not the same.Therefore,it is necessary to select the data processing mode and select the model structure before using the model for prediction.The empirical results show that the asset price forecasting based on the support vector machine can achieve a good prediction result,which has certain reference significance for investment practice.