Graph-Based Adaptive Fusion Network for Multi-Horizon Cryptocurrency Price Forecasting
6 affiliations
- 1Department of Computer Science, Faculty of Informatics, Mahasarakham University, Maha Sarakham 44150, Thailand
- 2Department of Information Technology, Faculty of Informatics, Mahasarakham University, Maha Sarakham 44150, Thailand
- 3Department of Geo-Informatics, Faculty of Informatics, Mahasarakham University, Maha Sarakham 44150, Thailand
- 4Faculty of Informatics, Mahasarakham University, Maha Sarakham 44150, Thailand
- 5Extend IT Resource Co., Ltd., Muang Thai-Phatra Complex Tower B, Huai Khwang, Bangkok 10310, Thailand
- 6Applied Statistics Research Unit, Department of Mathematics, Mahasarakham University, Maha Sarakham, 44150 Thailand
Abstract
This study presents a Graph-Based Adaptive Fusion Network (GAFN) for forecasting the closing prices of Bitcoin, Ethereum, Litecoin, and XRP at 7-, 14-, and 30-day horizons. GAFN combines a BiLSTM encoder, a graph attention network constructed from training-period return correlations, and target-conditioned adaptive fusion. It predicts standardized future log returns as intermediate outputs, which are converted to closing-price forecasts for evaluation. The experiments used 3131 aligned daily observations from Yahoo Finance covering 9 November 2017 to 5 June 2026. The data were divided chronologically into training, validation, and confirmatory test partitions. The final 156 observations, covering 1 January to 5 June 2026, yielded 97 forecast origins after sequence generation. Feature and target standardization and graph construction used the training partition only; model selection and residual scaling used the validation partition. GAFN was compared with Persistence, recurrent, fusion, attention-based, and graph-based models using asset-level NRMSE and an unweighted macro-average across the four assets. Each trainable model was evaluated over five random seeds. Several simpler models achieved lower macro-averaged NRMSE than GAFN at 7 days. At 14 days, GAFN tied with LSTM at the reported precision. At the 30-day horizon, GAFN achieved a macro-averaged test NRMSE of 0.03997 ± 0.00019, compared with 0.03979 for Persistence and 0.04062 for Concatenation Fusion. GAFN had the lowest mean NRMSE among the trainable models at this horizon, but the observed differences were not statistically significant after adjustment for multiple comparisons. The ablation results also showed no consistent improvement from any individual component across all three horizons. GAFN was therefore competitive in this experiment, but the results do not establish superiority beyond the tested assets and period.
Related
- MSAFP-Net: An Adaptive Fusion-Driven Multi-Scale Neural Network for Peak Production Forecasting in Gas Wells
- State-aware adaptive multimodal fusion for ultra-short-term solar irradiance forecasting using sky cloud images
- Hankel-Koopman Finite-Horizon Energy Decomposition of Coupled Experimental Data: A Three-Phase Data-Driven Twin Forecasting Framework
- Monotone-Constrained Diffusion Models for Long-Horizon Production Forecasting
- PREF-VG: Topology-Conditioned Residual Adaptation for Probabilistic Multi-Horizon Student Engagement Forecasting
BibTeX
@article{polpinij2026graphbased,
title = {{Graph-Based Adaptive Fusion Network for Multi-Horizon Cryptocurrency Price Forecasting}},
author = {Jantima Polpinij and Anirut Chotthanom and Theeraya Uttha and Woraphot Watthusin and Panu Tansorajprasert and Orawich Kumphon},
journal = {Mathematics},
year = {2026},
month = nov,
doi = {10.3390/math14214099},
url = {https://www.mdpi.com/2227-7390/14/21/4099},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Graph-Based Adaptive Fusion Network for Multi-Horizon Cryptocurrency Price Forecasting
AU - Jantima Polpinij
AU - Anirut Chotthanom
AU - Theeraya Uttha
AU - Woraphot Watthusin
AU - Panu Tansorajprasert
AU - Orawich Kumphon
PY - 2026
DA - 2026/11/08
JO - Mathematics
DO - 10.3390/math14214099
UR - https://www.mdpi.com/2227-7390/14/21/4099
AB - This study presents a Graph-Based Adaptive Fusion Network (GAFN) for forecasting the closing prices of Bitcoin, Ethereum, Litecoin, and XRP at 7-, 14-, and 30-day horizons. GAFN combines a BiLSTM encoder, a graph attention network constructed from training-period return correlations, and target-conditioned adaptive fusion. It predicts standardized future log returns as intermediate outputs, which are converted to closing-price forecasts for evaluation. The experiments used 3131 aligned daily observations from Yahoo Finance covering 9 November 2017 to 5 June 2026. The data were divided chronologically into training, validation, and confirmatory test partitions. The final 156 observations, covering 1 January to 5 June 2026, yielded 97 forecast origins after sequence generation. Feature and target standardization and graph construction used the training partition only; model selection and residual scaling used the validation partition. GAFN was compared with Persistence, recurrent, fusion, attention-based, and graph-based models using asset-level NRMSE and an unweighted macro-average across the four assets. Each trainable model was evaluated over five random seeds. Several simpler models achieved lower macro-averaged NRMSE than GAFN at 7 days. At 14 days, GAFN tied with LSTM at the reported precision. At the 30-day horizon, GAFN achieved a macro-averaged test NRMSE of 0.03997 ± 0.00019, compared with 0.03979 for Persistence and 0.04062 for Concatenation Fusion. GAFN had the lowest mean NRMSE among the trainable models at this horizon, but the observed differences were not statistically significant after adjustment for multiple comparisons. The ablation results also showed no consistent improvement from any individual component across all three horizons. GAFN was therefore competitive in this experiment, but the results do not establish superiority beyond the tested assets and period.
SN - 2227-7390
ER -