Research On The Method Of Forecasting The Fluctuation Trend Of Stock Returns

Stock is a kind of voucher for a securities,it can carry out market circulation and bring in profits.The research of the stock market is welcomed by scholars in the financial.This thesis analyzes the horizontal characteristics and volatility characteristics of the leading stock daily return rate,and performs short-term forecasting and volatility fitting of the yield.Time series analysis is a very important tool for studying financial data,so the thesis uses time series analysis method to study stock daily return rate and use SAS software programming to process data and build models.Firstly we use SAS software to fit ARMA model,ARIMA model,sparse coefficient model,ARCH model,GARCH model,ARGARCH model and other models,and uses AIC and BIC criteria and error square sum to optimize the model and predict.Then we consider the time series model,which consider the relationship between the stock daily return rate sequence and stock daily return rate series and the date.We also fit the multivariate time series ARIMAX model between the stock daily return rate series and the sequence itself.We also combine the heteroscedastic model and the multivariate time series model to fit the GARCHX model.Discovering that the optimal forecasting model for the stock return rate of the leading stocks is a multivariate time series model.This model fits the stock daily yield very well.We also estimate the risk interval,which can provide the reference for short-term investors and stock decisions.