| dc.contributor.advisor | Setiawaty, Berlian | |
| dc.contributor.advisor | Budiarti, Retno | |
| dc.contributor.author | Syamsudin, Smaragdy Radhiya | |
| dc.date.accessioned | 2025-12-23T08:45:45Z | |
| dc.date.available | 2025-12-23T08:45:45Z | |
| dc.date.issued | 2025 | |
| dc.identifier.uri | http://repository.ipb.ac.id/handle/123456789/171814 | |
| dc.description.abstract | Penelitian ini bertujuan untuk mengidentifikasi penggunaan model ARMA–
GARCH dalam memprediksi volatilitas Bitcoin serta mengevaluasi akurasinya
berdasarkan MAPE. Data harga harian Bitcoin periode 1 Januari 2023–31
Desember 2023 dipraproses melalui differencing satu kali, translasi agar seluruh
nilai positif, dan transformasi Box–Cox. Berdasarkan analisis ACF–PACF dan
perbandingan AIC, model mean terbaik adalah ARMA(0,0). Uji Ljung–Box
menunjukkan residual tidak berkorelasi, sedangkan uji ARCH–LM
mengindikasikan heteroskedastisitas bersyarat sehingga volatilitas dimodelkan
menggunakan GARCH(0,1). Evaluasi menunjukkan MAPE 20,223% pada data
train, 22,930% pada data uji 15 langkah ke depan, dan rata-rata MAPE 20,177%
dari lima percobaan set.seed terbaik, yang termasuk kategori prediksi baik. Hasil
ini menegaskan bahwa model ARMA–GARCH mampu menangkap pola harga dan
volatilitas Bitcoin secara stabil serta layak digunakan untuk peramalan jangka
pendek dan pendukung pengelolaan risiko pada aset kripto yang bergejolak.
Kata kunci: ARMA-GARCH, volatilitas, Bitcoin | |
| dc.description.abstract | This study aims to identify the use of the ARMA–GARCH model for predicting
Bitcoin volatility and to evaluate its accuracy using the MAPE metric. Daily Bitcoin
prices from January 1, 2023 to December 31, 2023 were preprocessed through onetime differencing, a translation to ensure positive values, and a Box–Cox
transformation. Based on ACF–PACF analysis and AIC comparison, the selected
mean model is ARMA(0,0). The Ljung–Box test indicates uncorrelated residuals,
while the ARCH–LM test confirms conditional heteroskedasticity, leading
volatility to be modeled using GARCH(0,1). The evaluation shows a train MAPE
of 20.223%, a 15-step-ahead test MAPE of 22.930%, and an average MAPE of
20.177% across the five best set.seed runs, classified as good prediction accuracy.
These findings demonstrate that the ARMA–GARCH model effectively captures
Bitcoin price dynamics and volatility, making it suitable for short-term forecasting
and supporting risk management in highly volatile crypto markets.
Keywords: ARMA-GARCH, volatility, Bitcoin | |
| dc.description.sponsorship | | |
| dc.language.iso | id | |
| dc.publisher | IPB University | id |
| dc.title | Peramalan Harga Bitcoin menggunakan Model ARMA-GARCH | id |
| dc.title.alternative | | |
| dc.type | Skripsi | |
| dc.subject.keyword | bitcoin | id |
| dc.subject.keyword | ARMA-GARCH | id |