Integrating Seasonal Trends and Market Volatility: A Hybrid SARIMA-XGBoost Model for Cryptocurrency Forecasting

Year 2025, Volume: 18 Issue: 2, 581 - 602, 31.08.2025

Melikşah Aydın , Eyyüp Ensari Şahin

https://doi.org/10.17218/hititsbd.1613531

Abstract

This paper explores the application of a hybrid SARIMA-XGBoost model for forecasting cryptocurrency prices within the context of financial time series analysis. Due to their dynamic nature, high volatility, seasonal patterns, and asymmetric market behavior, cryptocurrencies pose significant challenges for traditional forecasting methods. The study combines the strength of the SARIMA model in capturing seasonal and trend components with the XGBoost algorithm's ability to model nonlinear relationships. This hybrid approach aims to better represent the complex and non-stationary nature of cryptocurrency markets, thereby improving forecasting accuracy. Cryptocurrencies, such as Bitcoin, exhibit volatile price movements influenced by factors like periodic "halving" effects and volatility clustering. Additionally, investor behaviors, herd psychology, and social media sentiment contribute to the difficulty of price forecasting. In this study, residuals from the SARIMA model are used as inputs for the XGBoost algorithm to capture nonlinear patterns, effectively integrating both linear and nonlinear elements in a single model. The paper analyzes the model's performance using BTC (Bitcoin), XRP (Ripple), and ETH (Ethereum) price data. Metrics such as the Akaike Information Criterion (AIC), Mean Squared Error (MSE), and Mean Absolute Error (MAE) are employed to evaluate the model's accuracy. Results indicate that the SARIMA-XGBoost hybrid model outperforms standalone SARIMA and XGBoost models, achieving higher accuracy, especially during periods of market volatility. The hybrid model successfully captures the dynamic nature of cryptocurrency markets, producing lower error rates compared to other models. The findings demonstrate the effectiveness of hybrid modeling approaches in forecasting cryptocurrency prices and highlight the potential of optimized hybrid models in investment decision-making and risk management. However, the study also notes that the model's performance could be further enhanced during short-term fluctuations and highly volatile market conditions.

Keywords

SARIMA , XGBoost , Cryptocurrency Forecasting , Financial Time Series , Volatility , Hybrit Modeling

Mevsimsel Trendler ve Piyasa Volatilitesinin Entegrasyonu: Kripto Para Tahmini için Hibrit SARIMA-XGBoost Modeli

Abstract

Bu makale, finansal zaman serisi analizi bağlamında, kripto para birimlerinin fiyat tahmini için hibrit bir model olan SARIMA-XGBoost yöntemini ele almaktadır. Kripto para birimlerinin dinamik yapısı, yüksek volatilite, mevsimsel desenler ve asimetrik piyasa davranışları nedeniyle geleneksel yöntemlerle analiz edilmesi oldukça zordur. Makalede, SARIMA modelinin mevsimsel ve trend bileşenlerini ele almadaki gücü ile XGBoost algoritmasının doğrusal olmayan ilişkileri modelleme yeteneği birleştirilmiştir. Bu hibrit model, kripto para piyasasının karmaşık ve durağan olmayan yapısını daha iyi temsil ederek tahmin performansını artırmayı amaçlamaktadır. Kripto varlıklar, Bitcoin gibi önde gelen para birimlerinin dönemsel “yarılanma” etkileri ve volatilite kümeleşmesi gibi faktörler nedeniyle dalgalı fiyat hareketleri sergilemektedir. Aynı zamanda, yatırımcı davranışları, sürü psikolojisi ve sosyal medya etkisi gibi değişkenler de fiyat tahminlerini zorlaştırmaktadır. Bu çalışmada, SARIMA modelinin artık değerleri, doğrusal olmayan desenlerin modellenmesi için XGBoost’a girdi olarak verilmiştir. Böylece, hem doğrusal hem de doğrusal olmayan desenler aynı modelde bütünleştirilmiştir. Makale, BTC (Bitcoin), XRP (Ripple) ve ETH (Ethereum) fiyat verileri üzerinde modelin performansını analiz etmektedir. Model performansı, Akaike Bilgi Kriteri (AIC), Ortalama Kare Hata (MSE) ve Ortalama Mutlak Hata (MAE) ölçütleriyle değerlendirilmiştir. Sonuçlar, SARIMA-XGBoost modelinin bağımsız SARIMA ve XGBoost modellerine kıyasla daha yüksek doğruluk sağladığını göstermektedir. Özellikle volatil dönemlerde hibrit model, kripto para piyasasının dinamiklerini daha iyi yakalayarak daha küçük hata oranları üretmektedir. Bu çalışmanın bulguları, kripto para birimlerinin gelecekteki fiyatlarının tahmin edilmesi için hibrit modellemelerin etkili bir yaklaşım olduğunu kanıtlamaktadır. Aynı zamanda, optimize edilmiş hibrit modellerin yatırım kararları ve risk yönetimi süreçlerinde kullanılabilirliği vurgulanmaktadır. Ancak, kısa vadeli dalgalanmalar ve aşırı volatilite dönemlerinde modelin performansında iyileştirmeler yapılabileceği belirtilmektedir.

Keywords

SARIMA , XGBoost , Kripto Para Tahmini , Finansal Zaman Serisi , Volatilite , Hibrit Modelleme

References

• Achmadi, G. R., Saikhu, A., & Amaliah, B. (2023). Cryptocurrency price movement prediction using the hybrid sarımax-lstm method. 2023 International Conference on Advanced Mechatronics, Intelligent Manufacture and Industrial Automation (ICAMIMIA), 711-716. https://doi.org/10.1109/ICAMIMIA60881.2023.10427973
• Alessandretti, L., ElBahrawy, A., Aiello, L. M., & Baronchelli, A. (2018). Machine learning the cryptocurrency market. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3183792
• Alhnaity, B., & Abbod, M. (2020). A new hybrid financial time series prediction model. Engineering Applications of Artificial Intelligence, 95, 103873. https://doi.org/10.1016/j.engappai.2020.103873
• Bitto, A. K., Mahmud, I., Bijoy, M. H. I., Jannat, F. T., Arman, M. S., Shohug, M. M. H., ... & Jahan, H. (2022). Cryptoar: Scrutinizing the trend and market of cryptocurrency using machine learning approach on time series data. Indonesian Journal of Electrical Engineering and Computer Science, 28(3), 1684. https://doi.org/10.11591/ijeecs.v28.i3.pp1684-1696
• Bouri, E., Gupta, R., & Roubaud, D. (2019). Herding behaviour in cryptocurrencies. Finance Research Letters, 29, 216-221. https://doi.org/10.1016/j.frl.2018.07.008
• Bülbül, M. A. (2024). Hybrid optimal time series modeling for cryptocurrency price prediction: feature selection, structure and hyperparameter optimization. Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, 13(3), 731-743. https://doi.org/10.17798/bitlisfen.1479725
• Box, G. E. P., Jenkins, G. M., Reinsel, G. C., & Ljung, G. M. (2015). Time series analysis: Forecasting and control (5th ed.). Wiley. https://doi.org/10.1002/9781118619193
• Caporale, G., Kang, W., Spagnolo, F., & Spagnolo, N. (2019). Non-linearities, cyber attacks and cryptocurrencies. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3409138
• 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
• Derbentsev, V., Kibalnyk, L., & Radzihovska, Y. (2019). Modelling multifractal properties of cryptocurrency market. Periodicals of Engineering and Natural Sciences (Pen), 7(2), 690. https://doi.org/10.21533/pen.v7i2.559
• Fleischer, J., Laszewski, G., Theran, C., & Bautista, Y. (2022). Time series analysis of cryptocurrency prices using long short-term memory. Algorithms, 15(7), 230. https://doi.org/10.3390/a15070230
• Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5), 1189–1232. https://doi.org/10.1214/aos/1013203451
• Gao, M., Yang, H., Xiao, Q., & Goh, M. (2022). Covid-19 lockdowns and air quality: Evidence from grey spatiotemporal forecasts. Socio-Economic Planning Sciences, 83, 101228. https://doi.org/10.1016/j.seps.2022.101228
• Gupta, H., & Chaudhary, R. (2022). An empirical study of volatility in cryptocurrency market. Journal of Risk and Financial Management, 15(11), 513. https://doi.org/10.3390/jrfm15110513
• Haykır, Ö., & Yağlı, İ. (2022). Speculative bubbles and herding in cryptocurrencies. Financial Innovation, 8(1). https://doi.org/10.1186/s40854-022-00383-0
• Hoa, N., Thi, Q., & Ngoan, N. (2023). Time series prediction based on machine learning: A case study, temperature forecasting in Vietnam. Journal of Military Science and Technology, 85, 152-162. https://doi.org/10.54939/1859-1043.j.mst.85.2023.152-162
• Hossain, M., Ismail, M., & Hossain, M. (2022). Enhancing stock price prediction using empirical mode decomposition, rolling forecast, and combining statistical methods. International Journal of Computing and Digital Systems, 12(6), 1343-1356. https://doi.org/10.12785/ijcds/1201108
• Huang, H. (2023). Long-term changes in Ethereum prices: A normalized pandemic framework. BCP Business & Management, 38, 72-78. https://doi.org/10.54691/bcpbm.v38i.3672
• Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: Principles and practice (2nd ed.). OTexts. https://doi.org/10.1515/9783110612500
• Iqbal, M., Iqbal, M. S., Jaskani, F. H., Iqbal, K., & Hassan, A. (2021). Time-series prediction of cryptocurrency market using machine learning techniques. EAI Endorsed Transactions on Creative Technologies, 8(28), 170286. https://doi.org/10.4108/eai.7-7-2021.170286
• Jabeur, S. B., Khalfaoui, R., & Arfi, W. B. (2021). The effect of green energy, global environmental indexes, and stock markets in predicting oil price crashes: Evidence from explainable machine learning. Journal of Environmental Management, 298, 113511. https://doi.org/10.1016/j.jenvman.2021.113511
• Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., ... & Liu, T. Y. (2017). LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30, 3146–3154. https://doi.org/10.5555/3294996.3295074
• Kim, J., Won, J., Kim, H., & Heo, J. (2021). Machine-learning-based prediction of land prices in Seoul, South Korea. Sustainability, 13(23), 13088. https://doi.org/10.3390/su132313088
• Li, Z., Han, J., & Song, Y. (2020). On the forecasting of high‐frequency financial time series based on ARIMA model improved by deep learning. Journal of Forecasting, 39(7), 1081-1097. https://doi.org/10.1002/for.2677
• Liang, L. (2024). ARIMA with Attention-based CNN-LSTM and XGBoost hybrid model for stock prediction in the US stock market. SHS Web of Conferences, 196, 02001. https://doi.org/10.1051/shsconf/202419602001
• Liu, Z., Zhu, Z., Gao, J., & Xu, C. (2021). Forecast methods for time series data: A survey. IEEE Access, 9, 91896-91912. https://doi.org/10.1109/access.2021.3091162
• Nichani, R., Gasmi, L., Laiche, N., & Kabou, S. (2023). Optimizing financial time series predictions with hybrid ARIMA, LSTM, and XGBoost Models. Studies in Engineering and Exact Sciences. https://doi.org/10.54021/seesv5n2-582ojs.studiespublicacoes.com.br
• Papadimitriou, T., Gogas, P., & Athanasiou, A. (2022). Forecasting Bitcoin spikes: A GARCH-SVM approach. Forecasting, 4(4), 752-766. https://doi.org/10.3390/forecast4040041
• Rampini, L., & Cecconi, F. R. (2021). Artificial intelligence algorithms to predict Italian real estate market prices. Journal of Property Investment & Finance, 40(6), 588-611. https://doi.org/10.1108/jpif-08-2021-0073
• Song, X. (2023). Predicting cryptocurrency investment suitability using machine learning techniques. Applied and Computational Engineering, 29(1), 133-141. https://doi.org/10.54254/2755-2721/29/20230726
• Shen, Y., & Wang, H. (2023). Valuation and forecasting of cryptocurrency: Analysis of Bitcoin, Ethereum, and Dogecoin. BCP Business & Management, 38, 1067-1074. https://doi.org/10.54691/bcpbm.v38i.3828
• Takaishi, T. (2018). Statistical properties and multifractality of Bitcoin. Physica A: Statistical Mechanics and Its Applications, 506, 507-519. https://doi.org/10.1016/j.physa.2018.04.046
• Takaishi, T. (2021). Time-varying properties of asymmetric volatility and multifractality in Bitcoin. PLoS One, 16(2), e0246209. https://doi.org/10.1371/journal.pone.0246209
• Valencia, F., Gómez-Espinosa, A., & Valdés-Aguirre, B. (2019). Price movement prediction of cryptocurrencies using sentiment analysis and machine learning. Entropy, 21(6), 589. https://doi.org/10.3390/e21060589
• Zheng, M., Guo, F., Zhang, X., & Chang, C. (2023). The transaction behavior of cryptocurrency and electricity consumption. Financial Innovation, 9(1). https://doi.org/10.1186/s40854-023-00449-7