Hidden Markov Model untuk Deteksi Rezim dan Prediksi Kejatuhan Harga pada Bitcoin: Evaluasi Empiris Regime-Aware Stacking Ensemble

Bagus Satria Putra; Kumara Ari Yuana

doi:10.55681/armada.v4i10.4073

Authors

Bagus Satria Putra Program Studi Magister Teknik Informatika, Universitas Amikom Yogyakarta, Indonesia
Kumara Ari Yuana Fakultas Ilmu Komputer, Universitas Amikom Yogyakarta, Indonesia

DOI:

https://doi.org/10.55681/armada.v4i10.4073

Keywords:

Bitcoin, Peramalan Crash, Hidden Markov Model, Manajemen Risiko Investor, SHAP

Abstract

Ayunan harga Bitcoin yang keras membuat kemungkinan keruntuhan mendadak membayangi investor, namun sebagian besar skema peramalan mematok kaitan antara prediktor dan peluang crash sebagai sesuatu yang tetap tanpa memandang kondisi pasar. Studi ini menguji apakah pengondisian prediktor pada regime pasar laten mempertajam peramalan crash Bitcoin. Alur regime-aware menyatukan Hidden Markov Model Gaussian tiga state untuk deteksi regime, stacking ensemble yang memasangkan XGBoost dengan jaringan Long Short-Term Memory untuk klasifikasi, dan SHAP untuk interpretasi per regime, dievaluasi pada 2.884 catatan harian 2018–2025 dengan tujuh belas fitur teknikal, on-chain, dan sentimen. Crash didefinisikan sebagai penurunan kumulatif di atas lima belas persen dalam tujuh hari, dengan ambang sepuluh persen untuk uji sensitivitas. Kejadian crash berbeda tajam antar regime Bull, Bear, dan Sideways (Kruskal-Wallis p = 0,000289; rasio odds Bear terhadap Bull = 1,70). Namun tak ada rancangan regime-aware yang mengalahkan baseline buta regime: pada kondisi berdaya memadai, varian terbaik hanya menambah 0,0092 ± 0,0017 (rata-rata tiga eksekusi) pada AUPRC dengan Wilcoxon p pada rentang 0,81–1,00, dan XGBoost tunggal mengungguli seluruh varian bertumpuk. Deteksi regime terbukti sebagai pengukur diagnostik risiko, bukan penambah daya prediksi.

Downloads

Download data is not yet available.

References

Agakishiev, I., Härdle, W. K., Becker, D., & Zuo, X. (2025). Regime switching forecasting for cryptocurrencies. Digital Finance, 7, 107–131. https://doi.org/10.1007/s42521-024-00123-2

Babalos, V., Bouri, E., & Gupta, R. (2025). Does the introduction of US spot Bitcoin ETFs affect spot returns and volatility of major cryptocurrencies? The Quarterly Review of Economics and Finance, 102, 102006. https://doi.org/10.1016/j.qref.2025.102006

Baker, M., & Wurgler, J. (2007). Investor sentiment in the stock market. Journal of Economic Perspectives, 21(2), 129–152. https://doi.org/10.1257/jep.21.2.129

Berger, T., & Koubová, J. (2024). Forecasting Bitcoin returns: Econometric time series analysis vs. machine learning. Journal of Forecasting, 43(7), 2904–2916. https://doi.org/10.1002/for.3165

Bikhchandani, S., & Sharma, S. (2001). Herd behavior in financial markets. IMF Staff Papers, 47(3), 279–310.

Chandra, R., & He, Y. (2024). Review of deep learning models for crypto price prediction: Implementation and evaluation. arXiv. https://arxiv.org/abs/2405.11431

Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794. https://doi.org/10.1145/2939672.2939785

Davis, J., & Goadrich, M. (2006). The relationship between Precision-Recall and ROC curves. Proceedings of the 23rd International Conference on Machine Learning, 233–240. https://doi.org/10.1145/1143844.1143874

Dubey, R., & Enke, D. (2025). Bitcoin price direction prediction using on-chain data and feature selection. Machine Learning with Applications, 100674. https://doi.org/10.1016/j.mlwa.2025.100674

Giudici, P., & Polinesi, G. (2021). A hidden Markov model to detect regime changes in cryptoasset markets. Quality and Reliability Engineering International, 37(7), 3021–3033. https://doi.org/10.1002/qre.2960

Hastie, T., Tibshirani, R., & Friedman, J. (2009). The elements of statistical learning: Data mining, inference, and prediction (2nd ed.). Springer. https://doi.org/10.1007/978-0-387-84858-7

He, H., & Garcia, E. A. (2009). Learning from imbalanced data. IEEE Transactions on Knowledge and Data Engineering, 21(9), 1263–1284. https://doi.org/10.1109/TKDE.2008.239

Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735

Ji, S., Kim, J., & Im, H. (2019). A comparative study of Bitcoin price prediction using deep learning. Mathematics, 7(10), 898. https://doi.org/10.3390/math7100898

Köse, M. (2024). Evolving strategies in machine learning: A systematic review of concept drift detection. Information, 15(12), 786. https://doi.org/10.3390/info15120786

Liu, Z., Shu, M., & Zhu, W. (2024). Contrastive learning framework for Bitcoin crash prediction. Stats, 7(2), 402–433. https://doi.org/10.3390/stats7020025

Lundberg, S. M., & Lee, S. I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774.

Lundberg, S. M., Erion, G. G., & Lee, S. I. (2018). Consistent individualized feature attribution for tree ensembles. arXiv. https://arxiv.org/abs/1802.03888

Machimbo, E. W., Cheruiyot, W., Mocheche, G. M., & Mutegi, J. (2025). Applications of hidden Markov models in detecting regime changes in Bitcoin markets. Asian Journal of Probability and Statistics, 27(7), 127–138. https://doi.org/10.9734/ajpas/2025/v27i7781

Minsky, H. P. (1992). The financial instability hypothesis (Working Paper No. 74). The Jerome Levy Economics Institute.

Pakštaitė, V., Filatovas, E., Juodis, M., & Paulavičius, R. (2025). Bitcoin price regime shifts: A Bayesian MCMC and hidden Markov model analysis of macroeconomic influence. Mathematics, 13(10), Article 1577. https://doi.org/10.3390/math13101577

Park, S., & Yang, J. S. (2024). Machine learning models based on bubble analysis for Bitcoin market crash prediction. Engineering Applications of Artificial Intelligence, 135, Article 108857. https://doi.org/10.1016/j.engappai.2024.108857

Pennoni, F., Bartolucci, F., Ferrante, L., & Pandolfi, S. (2022). Exploring the dependencies among main cryptocurrency log-returns: A hidden Markov model. Economic Notes, 51(1), e12193. https://doi.org/10.1111/ecno.12193

Pennoni, F., Bartolucci, F., Forte, G., & Ametrano, F. (2022). Exploring the dependencies among main cryptocurrency log-returns: A hidden Markov model. Economic Notes, 51(1), Article e12193. https://doi.org/10.1111/ecno.12193

Rabiner, L. R. (1989). A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE, 77(2), 257–286. https://doi.org/10.1109/5.18626

Saito, T., & Rehmsmeier, M. (2015). The Precision-Recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLOS ONE, 10(3), e0118432. https://doi.org/10.1371/journal.pone.0118432

Santiment Academy. (2023). MVRV ratio: On-chain valuation metrics for cryptocurrency. Santiment.

Shu, M., & Zhu, W. (2024). Machine learning models based on bubble analysis for Bitcoin market crash prediction. arXiv. https://arxiv.org/abs/2312.00268

Undang-Undang Republik Indonesia Nomor 4 Tahun 2022 tentang Pengembangan dan Penguatan Sektor Keuangan (2022).