Advanced time series forecasting for commodities: Insights from the FEDformer model

Lei Ge, Qiwei Huang, Fengshuang Zhu and Shun Chen

Energy Economics, 2025, vol. 147, issue C

Abstract: Forecasting commodity prices is vital for economic policy, especially amid recent geopolitical tensions and market disruptions. In recent years, advanced deep learning models have become particularly effective in this domain. Among these models, RNN-based architectures like LSTM and GRU are known for their strong predictive capabilities. In this paper, we show that the FEDformer model offers superior accuracy in predicting commodity prices when compared not only to other deep learning approaches but also to a standard econometric baseline. The study applies these models to predict six commodity indices: the Bloomberg Commodity Index and its five component indices. The results show that the FEDformer model achieves a reduction in MAE of 38.5%, 56.6%, 34.6% and 29.2% compared to other RNN models and the econometric model. Furthermore, the Wilcoxon signed-rank test also indicates that the FEDformer significantly outperforms other RNN models across all metrics.

Keywords: Commodity price; Forecasting; Deep learning; FEDformer (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0140988325003378
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:eneeco:v:147:y:2025:i:c:s0140988325003378

DOI: 10.1016/j.eneco.2025.108513

Access Statistics for this article

Energy Economics is currently edited by R. S. J. Tol, Beng Ang, Lance Bachmeier, Perry Sadorsky, Ugur Soytas and J. P. Weyant

More articles in Energy Economics from Elsevier
Bibliographic data for series maintained by Catherine Liu ().