| The ARIMA model and the AR-GARCH model serve as time series models and both of them have their own characteristics.The full name of the ARIMA model is the Autoregressive Integrated Moving Average Model.It is the most common model used in time series prediction in statistical models as well as it is a linear model.AR-GARCH model takes into account the residual sequence's autocorrelation first.Then,fitting the autoregressive model to the residual sequence and fitting the GARCH model to the autoregressive residual sequence.The GARCH model is called the Generalized Autoregressive Conditional Heteroskedastic(GARCH)as well as it is a nonlinear model.In order to compare the prediction results of the two models in the different presence of heteroskedasticity,two columns of time series data of the international futures gold market with both linear and non-linear characteristics were selected.One is US gold continuous data,another is US gold 1204 data.Among them,there is a strong long-term heteroskedasticity in the continuous sequence of U.S.gold.However,U.S.Gold 1204 has short-term heteroscedasticity and heterogeneity is not very strong.Firstly,this article gives a brief overview of the method of time series analysis and selects the representative ARIMA model and AR-GARCH model as the research focus.Secondly,it elaborates the definition,advantages and disadvantages,modeling process and the application of the ARIMA model and the AR-GARCH model.Thirdly,it uses ARIMA model and AR-GARCH model to model the temporal data of the international futures gold market and make short-term predictions.Finally,the article analyzes the prediction results and compare the prediction results of the two models under different degrees.The results show that when there is a strong heteroskedasticity in the residual sequence,the AR-GARCH model has a better prediction effect for short-term prediction;When the heteroskedasticity of the residual sequence is weak,the ARIMA model has a better prediction effect. |
PDF Full Text Request