Peramalan Return Cryptocurrency Bitcoin Menggunakan Markov-Switching GARCH Long Short-Term Memory

Sianti, Radisha Fanni (2023) Peramalan Return Cryptocurrency Bitcoin Menggunakan Markov-Switching GARCH Long Short-Term Memory. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003211026-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (2MB) | Request a copy

Abstract

Perkembangan teknologi informasi telah memberikan perubahan besar pada aktivitas kehidupan masyarakat, termasuk dalam aktivitas ekonomi yang menghadirkan pembayaran non-tunai atau cashless. Pada dasarnya dalam penggunaan cashless dapat digunakan di berbagai opsi, salah satunya dalam melakukan transaksi cryptocurrency. Di Indonesia, cryptocurrency dianggap sebagai instrumen investasi yang diperdagangkan pada bursa berjangka dengan istilah Aset Kripto. Salah satu aset kripto yang populer saat ini adalah Bitcoin. Keterbatasan jumlah Bitcoin dan tidak adanya perantara pihak ketiga menyebabkan return Bitcoin memiliki volatilitas tinggi. Fluktuasi pergerakan return Bitcoin yang tinggi menjadi pertimbangan bagi investor untuk menentukan keputusan investasi yang tepat. Dalam konteks inklusi keuangan, penting untuk melakukan peramalan pada return Bitcoin untuk mengantisipasi fluktuasi tersebut. Pada penelitian ini dilakukan peramalan return Bitcoin menggunakan model Markov-Switching GARCH (MSGARCH) dengan kombinasi metode machine learning yaitu Long Short-Term Memory (LSTM). Model MSGARCH digunakan untuk memodelkan perubahan volatilitas dan memperhitungkan adanya sifat heteroskedastisitas pada return Bitcoin. Sedangkan model LSTM digunakan untuk menangkap ketergantungan jangka panjang antar variabel, informasi jangka pendek serta mengingat informasi penting di masa lalu. Hasil analisis menunjukkan bahwa model GARCH memiliki performa yang lebih baik dibandingkan model MSGARCH. Pendekatan model hybrid GARCH-LSTM dan MSGARCH-LSTM mampu menghasilkan performa yang lebih baik dibandingkan model GARCH dan MSGARCH. Model terbaik adalah model MSGARCH-LSTM dengan nilai HMAE sebesar 1,357 dan HMSE sebesar 3,915.
==================================================================================================================================
Information technology development has significantly changed people's activities, including economic activities that present non-cash or cashless payments. Cashless can be used in various options, one of which is making cryptocurrency transactions. In Indonesia, cryptocurrency is considered an investment instrument that is an agent on futures exchanges with the term Crypto Asset. One of the popular crypto assets today is Bitcoin. The limited number of Bitcoins and the absence of third-party intermediaries cause Bitcoin returns to have high volatility. Fluctuations in the movement of high Bitcoin returns are a consideration for investors to determine the right investment decision. In the context of financial inclusion, it is essential to forecast Bitcoin returns to anticipate these fluctuations. In this study, Bitcoin return forecasting was carried out using the Markov-Switching GARCH (MSGARCH) model with a combination of machine learning methods, namely Long Short-Term Memory (LSTM). The MSGARCH model is used to model volatility changes and consider heteroskedasticity in Bitcoin returns. Meanwhile, the LSTM model is used to capture long-term dependencies between variables and short-term information and remember important information from the past. The analysis results show that the GARCH model performs better than the MSGARCH model. The GARCH-LSTM and MSGARCH-LSTM hybrid model approaches can perform better than the GARCH and MSGARCH models. The best model is the MSGARCH-LSTM model, with an HMAE value of 1.357 and an HMSE of 3.915.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Bitcoin, GARCH, Keuangan, LSTM, MSGARCH; Bitcoin, Finance, GARCH, LSTM, MSGARCH.
Subjects:	Q Science > QA Mathematics > QA274.7 Markov processes--Mathematical models. Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Radisha Fanni Sianti
Date Deposited:	21 Sep 2023 04:07
Last Modified:	21 Sep 2023 04:07
URI:	http://repository.its.ac.id/id/eprint/104327

Actions (login required)

View Item