Abstract

Model Hybrid ARIMA-GARCH merupakan model penggabungan dari model ARIMA dan GARCH, yang dapat digunakan untuk mengatasi masalah residual model ARIMA yang terindikasi adanya heteroskedastik dalam variansi residual (volatilitas). Tujuan dari penelitian ini adalah untuk menemukan model terbaik Hybrid ARIMA-GARCH untuk data harga emas dan meramalkan data emas periode Juni sampai Oktober 2016. Metode penelitian meliputi perumusan masalah, pengumpulan data, pengolahan dan analisis data, serta penarikan kesimpulan. Tahapan dalam analisis dan pembahasan yaitu statistika deskriptif, pengujian stasioneritas, pembentukan model kondisional mean (ARIMA), pembentukan model kondisional varian (GARCH), penggabungan model hybrid ARIMA-GARCH, menentukan model terbaik hybrid ARIMA-GARCH, melakukan pengukuran akurasi peramalan hybrid ARIMA-GARCH, dan peramalan. Hasil dari penelitian ini diperoleh model terbaik untuk harga emas adalah hybrid ARIMA(2,1,3)-GARCH(1,1) dengan nilai MAPE = 2,2685% dan nilai MPE = -0,01543. Berdasarkan model terbaik tersebut diperoleh hasil peramalan untuk periode Juni sampai Oktober 2016 berturut–turut adalah Rp524.722,5276; Rp522.404,5077; Rp501.819,4615; Rp501.514,1764; Rp505.704,409, yang menunjukkan bahwa harga emas pada bulan Juni sampai dengan September 2016 mengalami penurunan harga.


 


Hybrid ARIMA-GARCH is the models to combining ARIMA models and GARCH models, whish can be use to estimate residual ARIMA models that indicate the existence of heteroskedasticity from residual’s variance (volatility). The purpose of this research was to find the best model hybrid ARIMA-GARCH for gold price and forecast gold price from June – October 2016. The method includes the formulation of the problem, data collection, processing and data analysis, then also conclusion. Stage in the analysis and discussion that is descriptive statistics, stationary test, the estimate of the conditional mean (ARIMA), the estimate conditional variance (GARCH), combining hybrid ARIMA-GARCH models, determine the best model of hybrid ARIMA-GARCH models, measurement accuracy of forecasting hybrid ARIMA-GARCH, and forecasting. The result of the research were obtained the best model for the price gold is a hybrid                    ARIMA(2,1,3)-GARCH(1,1) with the value of MAPE = 2,2685% and the value of MPE = -0,01543. Based on the best model, obtained the forecast result for the period from June-October 2016 in a row are Rp524.722,5276; Rp522.404,5077; Rp501.819,4615; Rp501.514,1764; Rp505.704,409 which shows that the gold price in June to the September 2016 experienced a decline in gold prices continuously.