Stock forecasting method based on wavelet analysis and ARIMA-SVR model

The people's living standards gradually increased, in addition to food and clothing problems, more and more people began to focus on the stock market investment, which also provides for the development of the stock market is a necessary condition. The stock market is not only the focus of investors, but also a valuable place for scholars to study and research. This paper presents a stock forecasting model based on wavelet analysis and ARIMA-SVR. The stock price is decomposed into reconstructed part and error part by wavelet decomposition and wavelet reconstruction. Then, the ARIMA model and the SVR model are used to forecast the reconstructed part and the error part respectively, and the final prediction results are combined to obtain the final prediction results. The daily closing price of Shanghai Pudong Development Bank from January 5, 2015 to January 29, 2016, was selected as the experimental data, with a total of 250, of which the first 220 were the training set and the last 30 were the test set to do simulation experiment. The experimental results show that compared with the single forecasting model, the proposed model is an effective method for forecasting stock price, which greatly improves the accuracy of forecasting.