aiqbench analysis
This paper introduces MoFE, a novel architecture combining Fourier Neural Operators with Mixture-of-Experts specifically designed to model the non-stationary dynamics of cryptocurrency markets. By treating volatility as multi-frequency components derived from stochastic differential equations, the framework effectively mitigates phase-lag issues common in conventional deep learning models for financial time series. The results demonstrate state-of-the-art directional accuracy and risk-adjusted returns on Bitcoin data spanning 2020 to 2025, offering a rigorous approach to function-to-function mapping in high-frequency trading environments.
MoFE: A Novel Mixture-of-Experts Framework with Fourier Neural Operators for Cryptocurrency Forecasting
arXiv:2608.17342v1 Announce Type: new Abstract: Forecasting cryptocurrency prices remains a formidable challenge due to inherent non-stationarity, abrupt regime shifts, and multi-scale stochastic dependencies. Conventional deep learning models often struggle to capture complex underlying dynamics, frequently resulting in persistent phase-lagged predictions. To address these limitations, we propose MoFE, a novel deep learning framework that integrates Fourier Neural Operators (FNOs) within a Mixture-of-Experts (MoE) architecture. Rooted in the theoretical framework of stochastic differential equations, MoFE conceptualizes cryptocurrency volatility as a superposition of multi-frequency components, which includes user network based fundamental growth, mining costs and halving mechanism caused seasonal volatility, and market sentiment-induced chaos. Specifically, specialized adaptive FNO (AFNO) and Convolution dual-domain experts learn continuous function-to-function mappings to encapsulate global spectral trends, cyclical adjustments and microstructures, while a dynamic gating based MoE mechanism enables adaptive strategy switching across diverse market regimes. Extensive experiments on Bitcoin datasets spanning January 2020 to December 2025 demonstrate that MoFE achieves state-of-the-art (SOTA) performance in both T+1 and T+5 forecasting horizons. Notably, the model effectively mitigates the phase-lag effect, delivering superior Directional Accuracy (DA) and Information Coefficient (IC). In high-fidelity simulated trading environments, these predictive gains transfer into significant excess returns and robust risk-adjusted performance, characterized by a high Sharpe ratio.
Abstract:Forecasting cryptocurrency prices remains a formidable challenge due to inherent non-stationarity, abrupt regime shifts, and multi-scale stochastic dependencies. Conventional deep learning models often struggle to capture complex underlying dynamics, frequently resulting in persistent phase-lagged predictions. To address these limitations, we propose MoFE, a novel deep learning framework that integrates Fourier Neural Operators (FNOs) within a Mixture-of-Experts (MoE) architecture. Rooted in the theoretical framework of stochastic differential equations, MoFE conceptualizes cryptocurrency volatility as a superposition of multi-frequency components, which includes user network based fundamental growth, mining costs and halving mechanism caused seasonal volatility, and market sentiment-induced chaos. Specifically, specialized adaptive FNO (AFNO) and Convolution dual-domain experts learn continuous function-to-function mappings to encapsulate global spectral trends, cyclical adjustments and microstructures, while a dynamic gating based MoE mechanism enables adaptive strategy switching across diverse market regimes. Extensive experiments on Bitcoin datasets spanning January 2020 to December 2025 demonstrate that MoFE achieves state-of-the-art (SOTA) performance in both T+1 and T+5 forecasting horizons. Notably, the model effectively mitigates the phase-lag effect, delivering superior Directional Accuracy (DA) and Information Coefficient (IC). In high-fidelity simulated trading environments, these predictive gains transfer into significant excess returns and robust risk-adjusted performance, characterized by a high Sharpe ratio.
9 pages, 7 figures. Published in 2026 IEEE International Conference on Blockchain and Cryptocurrency (ICBC)
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2608.17342 [cs.LG]
(or arXiv:2608.17342v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.17342 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
2026 IEEE International Conference on Blockchain and Cryptocurrency (ICBC), 2026, pp. 1-9
https://doi.org/10.1109/ICBC67748.2026.11575439 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 332 words
Related coverage
Fast Training of Mixture-of-Experts for Time Series Forecasting via Expert Loss Integration
arXiv · Sep 18, 2026
Localized Operator Learning with Adaptive Partition-of-Unity Mixture-of-Expert Networks
arXiv · Oct 6, 2026
Communication Dynamics Neural Networks: FFT-Diagonalized Layers for Improved Hessian Conditioning at Reduced Parameter Count
arXiv · Jun 10, 2026
CARNet: Channel-Adaptive Receiver Network for Robust NextG Communications
arXiv · Aug 4, 2026