Haber duyarlılığı ve teknik göstergeleri kullanan Bitcoin fiyat tahminine yönelik hibrit bir FinBERT-LSTM çerçevesi

Abstract

Bu çalışma, Bitcoin’in saatlik ve günlük kapanış fiyatlarını tahmin etmek amacıyla duygu analizi ile derin öğrenmeyi entegre eden hibrit bir tahmin çerçevesi önermektedir. Haziran 2021-Kasım 2025 dönemine ait saatlik BTC/USD verileri, aynı dönemde yayımlanmış yaklaşık 326.000 Bitcoin ilişkili haber başlığı ile birleştirilmiştir. Finansal metinler üzerinde eğitilmiş transformer tabanlı FinBERT modeli kullanılarak her bir haber için [-1, +1] aralığında duygu skorları üretilmiş ve bu skorlar işlem hacmi, MACD ve RSI gibi teknik göstergelerle entegre edilmiştir. Oluşturulan birleşik özellik seti, zamansal bağımlılıkları modellemek amacıyla LSTM ağı ile analiz edilmiştir. Ampirik sonuçlar, duygu destekli hibrit modellerin yalnızca teknik verilere dayalı modellere kıyasla RMSE, MAE, MAPE ve R² metriklerinde tutarlı biçimde üstün performans sağladığını göstermektedir. Saatlik hibrit model, 1.009 USD RMSE ve %99,23 R² değeri ile en yüksek performansı sergilemiştir. Ayrıca, 30 günlük örneklem dışı gerçek zamanlı testte 941 USD RMSE elde edilmiştir. Örneklem içi ve dışı sonuçların tutarlılığı, önerilen çerçevenin zamana karşı istikrarlı tahmin performansı sunduğunu ortaya koymaktadır.

Keywords

• Bitcoin fiyat tahmini
• kripto para piyasaları
• duygu analizi
• FinBERT
• derin öğrenme
• LSTM
• zaman serisi tahmini

A hybrid FinBERT-LSTM framework for Bitcoin price forecasting using news sentiment and technical indicators

Abstract

This study proposes a hybrid forecasting framework that integrates sentiment analysis with deep learning to predict Bitcoin’s hourly and daily closing prices. Hourly BTC/USD market data spanning June 2021 to November 2025 were combined with approximately 326,000 Bitcoin-related news headlines published over the same period. Sentiment scores in the range of [-1, +1] were generated for each headline using FinBERT, a transformer-based language model trained on financial texts, and were subsequently integrated with technical indicators such as trading volume, MACD, and RSI. The resulting combined feature set was modeled using an LSTM network to capture temporal dependencies. Empirical results demonstrate that sentiment-enhanced hybrid models consistently outperform models based solely on technical indicators across RMSE, MAE, MAPE, and R² metrics. The hourly hybrid model achieved the best performance, with an RMSE of 1,009 USD and an R² of 99.23%. Furthermore, a 30-day out-of-sample real-time evaluation yielded an RMSE of 941 USD. The consistency between in-sample and out-of-sample results indicates that the proposed framework maintains stable predictive performance over time.

Keywords

• Bitcoin price forecasting
• cryptocurrency markets
• sentiment analysis
• FinBERT
• deep learning
• LSTM
• time series forecasting

References

1. Popper, N. Digital Gold: The Untold Story of Bitcoin, Penguin Books, Westminster UK, (2015).
2. Yermack, D., Handbook of Digital Currency, Is Bitcoin a real currency? An economic appraisal, 31-43, Cambridge USA, (2013).
3. Glaser, F., Zimmermann, K., Haferkorn, M., Weber, M. C. and Siering M., Bitcoin-asset or currency? Revealing users’ hidden intentions, Twenty Second European Conference on Information Systems, 1–14, Tel Aviv, (2014).
4. Gandal, N., Hamrick J., Moore T. and Oberman, T., Price manipulation in the Bitcoin ecosystem, Journal of Monetary Economics, 95, 86–96, (2018).
5. Gyamerah, S. A., Two- Stage Hybrid Machine Learning Model for High- Frequency Intraday Bitcoin Price Prediction Based on Technical Indicators, Variational Mode Decomposition, and Support Vector Regression, Complexity, 1-15, (2021).
6. Gao, Z., He, Y., Kuruoglu E. E., A Hybrid Model Integrating LSTM and Garch for Bitcoin Price Prediction, 2021 IEEE 31st International Workshop on Machine Learning for Signal Processing (MLSP), Gold Coast, Australia, 1-6, (2021).
7. Araci, D., Finbert: Financial sentiment analysis with pre-trained language models. Master’s Thesis, University of Amsterdam, Information Studies: Data Science, Amsterdam, (2019).
8. Zhu, Y., Ma, J., Gu, F., Wang, J., Li, Z., Zhang, Y., Xu, J., Li, Y., Wang, Y., Yang, X., Price Prediction of Bitcoin Based on Adaptive Feature Selection and Model Optimization. Mathematics, 11, 1335, (2023).

9. Ladhari, A., Boubaker, H., Deep Learning Models for Bitcoin Prediction Using Hybrid Approaches with Gradient-Specific Optimization, Forecasting, 6, 279–295, (2025).
10. Bourday R., Aatouchi I., Kerroum M. A., Zaaouat A., Cryptocurrency Forecasting Using Deep Learning Models: A Comparative Analysis, HighTech and Information Journal, 5, 1055-1067, (2024).
11. Boadi E., Bitcoin Price Forecasting Based on Hybrid Variational Mode Decomposition and Long Short Term Memory Network, (2025).
12. Patel R., Chauhan J., Tiwari N. K., Upaddhyay V., Bajpai A., A Deep Learning Framework for Hourly Bitcoin Price Prediction Using Bi LSTM and Sentiment Analysis of Twitter Data, SN Computer Science, 5, 767, (2024).
13. Htay, H.S., Ghahremani,M., Shiaeles, S., Enhancing Bitcoin Price Prediction with DeepLearning: Integrating Social Media Sentiment and Historical Data, Applied Sciences, 15, 1554, (2025).
14. Gür Y. E., Ünal E., The Impact of News Sentiment on the Bitcoin Price via Machine Learning and Deep Learning- Based NLP Models, Journal of Forecasting, 1–29 (2025).
15. Lakhwan D., Dave A., Determining The Most Efficient Technical Indicator Of Investing In Financial Markets Based On Trends, Volume, Momentum And Volatility, Master’s Thesis, Lazarski University, Faculty of Economics and Management, Warsaw, (2021).
16. Egly H., Near Future Predictive Power of ARIMA and RSI on Large Capital OMXS30, Bachelors Thesis in Statistics, Örebro University, School of Business, Örebro, (2023).
17. Cohen G., Trading cryptocurrencies using algorithmic average truerange systems. Journal of Forecasting, 42, 212-222, (2023).
18. Chicco, D., Warrens, M. J., Jurman G., The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation, PeerJ Computer Science, 7:e623, (2021).
19. Neill, S. P., Hashemi M. R., Fundamentals of Ocean Renewable Energy in Neill, S. P., Ocean Modelling for Resource Characterization, Academic Press, 193-235, Cambridge MA, (2018).
20. Tatachar A., Comparative Assessment of Regression Models Based On Model Evaluation Metrics, International Research Journal of Engineering and Technology, 8, 9, 853-860, (2021).
21. Kilimci, H., Yıldırım, M., Kilimci, Z. H., The Prediction of Short-Term Bitcoin Dollar Rate (BTC/USDT) using Deep and Hybrid Deep Learning Techniques, 5th International Symposium on Multidisciplinary Studies and Innovative Technologies, 633-637, Ankara, Turkey, (2021).
22. Manaswi, N. K., Deep Learning with Applications Using Python Chatbots and Face, Object, and Speech Recognition With TensorFlow and Keras, Apress, Bangalore, Karnataka, India, (2018).