Peramalan Indeks Harga Saham dengan Autoregressive Moving Average Generelized Autoregressive Conditional Heteroscedasticity (ARMA-GARCH)

Abstract. The stock price index is the information the public needs to know the development of stock price movements. Stock price forecasting will provide a better basis for planning and decision making. The forecasting model that is often used to model financial and economic data is the Autoregressive Moving Average (ARMA). However, this model can only be used for data with the assumption of stationarity to variance (homoscedasticity), therefore an additional model is needed that can model data with heteroscedasticity conditions, namely the Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model. This study uses data partitioning in pre-pandemic conditions and during the pandemic, Insample data with pre-pandemic conditions and insample data during pandemic conditions. Based on the research results, the GARCH model (1,1) was obtained with the conditions before the pandemic and GARCH (1,2) during the pandemic condition. The forecasting model obtained has met the eligibility requirements of the GARCH model. If the forecasting model fulfills the eligibility requirements, then MAPE calculations are performed to see the accuracy of the forecasting model. And obtained MAPE in the conditions before the pandemic and during the pandemic in the very good category. Abstrak. Indeks harga saham merupakan informasi yang diperlukan masyarakat untuk mengetahui perkembangan pergerakan harga saham. Peramalan harga saham akan memberikan dasar yang lebih baik bagi perencanaan dan pengambilan keputusan. Model peramalan yang sering digunakan untuk memodelkan data keuangan dan ekonomi adalah Autoregrresive Moving Average (ARMA). Namun model tersebut hanya dapat digunakan untuk data dengan asumsi stasioneritas terhadap varian (homoskedastisitas), oleh karena itu diperlukan suatu model tambahan yang bisa memodelkan data dengan kondisi heteroskedastisitas, yaitu model Generalized Autoregressive Conditional Heteroscedastisity (GARCH). Penelitian ini menggunakan partisi data pada kondisi sebelum pandemi dan saat pandemi berlangsung data Insample dengan kondisi sebelum pandemi dan insample pada kondisi pandemi. Berdasarkan hasil penelitian, maka didapat model GARCH (1,1) dengan kondisi sebelum pandemi dan GARCH (1,2) saat kondisi pandemi. Model peramalan yang didapat sudah memenuhi syarat kelayakan model GARCH. Apabila model peramalan terpenuhi syarat kelayakannya maka dilakukan perhitungan MAPE untuk melihat keakuratan model peramalannya. Dan diperoleh MAPE pada kondisi sebelum pandemi dan saat pandemi dengan kategori sangat baik.