Search papers, labs, and topics across Lattice.
This paper introduces MoFE, a novel deep learning framework that combines Mixture-of-Experts (MoE) architecture with Fourier Neural Operators (FNOs) to enhance cryptocurrency price forecasting. By modeling volatility as a superposition of multi-frequency components and employing adaptive strategies for different market regimes, MoFE effectively addresses the challenges of non-stationarity and phase-lagged predictions. Extensive experiments demonstrate that MoFE achieves state-of-the-art performance in both T+1 and T+5 forecasting horizons, translating predictive improvements into significant excess returns in simulated trading environments.
MoFE not only mitigates phase-lag effects in cryptocurrency forecasting but also translates its predictive edge into substantial trading profits.
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.
