PriceFM: Foundation Model for Probabilistic
Electricity Price Forecasting

Runyao Yu^1,2,3, Chenhui Gu¹, Jochen Stiasny¹, Qingsong Wen⁴,

Wasim Sarwar Dilov³, Lianlian Qi^2,5, Jochen L. Cremer^1,2

¹ Delft University of Technology, ² Austrian Institute of Technology, ³ Rimac Technology,
⁴ Squirrel Ai Learning, ⁵ Technical University of Munich
Preprint

Code Paper

Example forecasting on testing samples

Abstract

We introduce a comprehensive and up-to-date dataset across 24 European countries (38 regions), spanning from 2022 to 2025. Building on this groundwork, we propose PriceFM, a spatiotemporal foundation model that integrates graph-based inductive biases to capture spatial interdependencies across interconnected electricity markets. The model is designed for multi-region, multi-timestep, and multi-quantile probabilistic electricity price forecasting. Extensive experiments and ablation studies confirm the model's effectiveness, consistently outperforming competitive baselines and highlighting the importance of spatial context in electricity markets.

Structure of PriceFM. The input (price and energy) features are passed into a fusion block to learn regional representation. These regional representations are then stacked to form the spatial representation. Next, the spatial representation is passed to the graph block to produce the decay-guided spatial representation. Finally, the decayed spatial representation is fed into hierarchical quantile heads to produce joint forecasts.

European-level energy data in 2024, averaged across regions. a Electricity price. Price spikes sharply during the morning and evening peak, dip around midday. b Forecasted load. Load exhibits a double-peak each day. c Forecasted solar power generation. Solar is zero overnight, rises in a smooth bell curve to a strong midday maximum, then falls back to zero by dusk. d Forecasted wind power generation (onshore and offshore). Wind lacks a daily pattern, fluctuates with high-frequency spikes.

Spatial distribution of electricity price and number of neighboring regions. a Electricity prices for 38 European regions averaged from 2022-01-01 to 2025-01-01. A significant zonal price difference is observed between north and south regions. b Number of neighboring regions that are directly connected to certain region via transmission lines. For example, France (FR) and Portugal (PT) are directly connected to Spain (ES), thus the number of neighboring regions for ES is 2. The mean value across all regions is 3.4.

Results of model comparison.

Results of ablation studies.

BibTeX

If you use our code or find our paper useful, please cite:

@misc{yu2025pricefm,
      title={PriceFM: Foundation Model for Probabilistic Electricity Price Forecasting}, 
      author={Runyao Yu and Chenhui Gu and Jochen Stiasny and Qingsong Wen and Wasim Sarwar Dilov and Lianlian Qi and Jochen L. Cremer},
      year={2025},
      eprint={2508.04875},
      archivePrefix={arXiv},
      primaryClass={cs.CE},
      url={https://arxiv.org/abs/2508.04875}, 
}