STUDI KOMPARASI MODEL PERAMALAN INDEKS SAHAM SMINFRA18 MENGGUNAKAN ARIMA DAN GARCH

This research aims to find out the comparison of forecasting models in predicting SMINFRA18. In this research there are 2 forecasting models, namely ARIMA and GARCH forecasting. The population in this study is the SMINFRA18 weekly closing value data for the period 3 March 2013 to 26 March 2023, totaling 522 time series data. The research results show that the best forecasting models for predicting the SMINFRA18 index are ARIMA (12,1,12) and GARCH (1.1). In the ARIMA model (12,1,12) there are assumptions that are not fulfilled, namely the assumption of homoscedasticity or in the model there is an element of heteroscedasticity so that the GARCH (1,1) model with a MAPE of 1.94% must be used as the best forecasting model to predict SMINFRA18.