Research On Methods Of Stock Trends Prediction Based On Hybrid Model

With the continuous development of Chinese financial market and the gradual improvement of the financial system,people's interest in investing in financial markets has become increasingly strong,but also accompanied by a strong demand for accurate and effective financial information services.Financial markets play a vital role in the country's economic system.In recent years,many researchers have begun to focus on the field and use a variety of computer-related technologies to try to solve the problem of financial data prediction.This paper designs a data set with a market representation for the prediction of financial varieties(stocks and indices),and uses the deep learning method and time series analysis method to provide a system platform to provide the financial market forecast information service.The main research contents of this article is divided into the following aspects:Construction and Preprocessing of Financial Timing Data Set.Because of the policy reasons,financial data is open,so there is no organization or individual publication standard research data set in the field of financial forecasting research.This paper is constructed by acquiring historical market data of representative stocks and indices in China A-share market.This data set is the basis of our research.Research on Financial Timing Prediction Based on Deep Learning and ARIMA Model.In this paper,the deep learning model and ARIMA model are designed according to the problem of financial timing forecasting,and the further analysis and optimization of the application of this problem can make it better to explore the change of financial data.Constructing an Information Service Platform for Forecasting Financial Trends with Hybird Model.Based on the financial time series data set,the deep learning model and the ARIMA model,this paper constructs an information service platform for financial trend forecasting with hybird model to analyze and forecast the future trends and show the historical performance of specific stocks and indices.The evaluation of system platform.In this paper,the validity of the method is verified,and the real value verification and the traditional method are compared.On the one hand,the results of the system prediction analysis are compared with the real market performance.On the other hand,the advantages and disadvantages of the system are further evaluated by comparing the research methods and the results of the other research which presented by some organizations and individuals.In contrast to the real market performance,the study of this subject has achieved good results of RMSE error.In contrast to the results obtained by traditional single-model research methods such as SVM or ARIMA,model approach closer to the real market performance,achieved significantly better than the effect of other methods.