![[2602.12380] TFT-ACB-XML: Decision-Level Integration of Customized Temporal Fusion Transformer and Attention-BiLSTM with XGBoost Meta-Learner for BTC Price Forecasting](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.12380] TFT-ACB-XML: Decision-Level Integration of Customized Temporal Fusion Transformer and Attention-BiLSTM with XGBoost Meta-Learner for BTC Price Forecasting
Summary
This article presents the TFT-ACB-XML framework, a hybrid model integrating Temporal Fusion Transformer and Attention-BiLSTM with XGBoost for accurate Bitcoin price forecasting.
Why It Matters
Accurate Bitcoin price forecasting is crucial for investors and traders due to the cryptocurrency's volatility. This research addresses existing model limitations by combining advanced deep learning techniques, potentially improving prediction accuracy and market understanding.
Key Takeaways
- The TFT-ACB-XML framework integrates multiple models for enhanced BTC price prediction.
- Utilizes a novel error-reciprocal weighting strategy to optimize model outputs.
- Demonstrates superior performance with a MAPE of 0.65% compared to existing models.
Computer Science > Machine Learning arXiv:2602.12380 (cs) [Submitted on 12 Feb 2026] Title:TFT-ACB-XML: Decision-Level Integration of Customized Temporal Fusion Transformer and Attention-BiLSTM with XGBoost Meta-Learner for BTC Price Forecasting Authors:Raiz Ud Din (1), Saddam Hussain Khan (2) ((1) Artificial Intelligence Lab, Department of Computer Systems Engineering, University of Engineering and Applied Sciences (UEAS), Swat, Pakistan, (2) Interdisciplinary Research Center for Smart Mobility and Logistics, King Fahad University of Petroleum and Minerals (KFUPM), Dhahran, Saudi Arabia) View a PDF of the paper titled TFT-ACB-XML: Decision-Level Integration of Customized Temporal Fusion Transformer and Attention-BiLSTM with XGBoost Meta-Learner for BTC Price Forecasting, by Raiz Ud Din (1) and 9 other authors View PDF Abstract:Accurate forecasting of Bitcoin (BTC) has always been a challenge because decentralized markets are non-linear, highly volatile, and have temporal irregularities. Existing deep learning models often struggle with interpretability and generalization across diverse market conditions. This research presents a hybrid stacked-generalization framework, TFT-ACB-XML, for BTC closing price prediction. The framework integrates two parallel base learners: a customized Temporal Fusion Transformer (TFT) and an Attention-Customized Bidirectional Long Short-Term Memory network (ACB), followed by an XGBoost regressor as the meta-learner. The customized TFT model ha...