← Back to all news
arXiv Machine Learning By Zhenghua Pan, Ahmed Aziz Ezzat

Evaluating Time Series Foundation Models for Electricity Price Forecasting: Contamination Risk, Distributional Shifts, and Covariate Dependence

Read the original on arXiv Machine Learning →

arXiv:2607. 02623v1 Announce Type: new Abstract: Time series foundation models (TSFMs) have shown strong zero-shot forecasting performance, but their generalization in covariate-driven, non-stationary settings is underexplored.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Statistics ML
Sep 3

Online Multivariate Regularized Distributional Regression for High-dimensional Probabilistic Electricity Price Forecasting

The paper presents an online algorithm for multivariate distributional regression tailored to high‑dimensional probabilistic electricity price forecasting. It models conditional means, variances, and dependence structures of 24‑hour day‑ahead prices using coordinate descent and LASSO regularization, enabling scalable estimation in large covariate spaces. A regularized estimation path allows early stopping to prevent overfitting, and the method delivers interpretable, well‑calibrated joint prediction intervals, outperforming benchmarks on German market data.

By Simon Hirsch
More like this →
arXiv Machine Learning
Jul 21

Time-Aware Prior Fitted Networks for Zero-Shot Forecasting with Exogenous Variables

arXiv:2603. 15802v2 Announce Type: replace Abstract: In many time series forecasting settings, the target time series is accompanied by exogenous covariates, such as promotions and prices in retail demand; temperature in energy load; calendar and holiday indicators for traffic or sales; and grid load or fuel costs in electricity pricing.

By Andres Potapczynski, Ravi Kiran Selvam, Tatiana Konstantinova, Malcolm Wolff, Kin G. Olivares, Ruijun Ma, Michael W. Mahoney, Andrew Gordon Wilson, Boris N. Oreshkin, Dmitry Efimov
benchmarks
More like this →
arXiv Machine Learning
Sep 2

Foundation models for electricity price forecasting and battery arbitrage: Can they replace market-specific forecasting models?

The study evaluates nine foundation model variants against two leading electricity price forecasting benchmarks across Germany, Poland, and Spain for 2021‑2025. Only the TabPFN models consistently outperform the benchmarks in both point and probabilistic accuracy, yet their economic advantage varies with bidding strategy and risk tolerance. The results indicate that foundation models cannot universally replace market‑specific models; their usefulness depends on the chosen architecture and the particular decision problem.

By Arkadiusz Lipiecki, Rafa{\l} Weron
benchmarks
More like this →
arXiv Machine Learning
Jul 3

Probabilistic Low-Voltage Peak Load Forecasting with Time Series Foundation Models Evaluated on Application-Oriented Metrics

arXiv:2607. 01966v1 Announce Type: new Abstract: Low-voltage load forecasting is an important component in current and future energy systems with a high degree of electrification and decentralized generation.

By Benedikt Kaas, Manuel Treutlein, Hannes Benedikt Gerber, Oliver Neumann, Cheewan Phatthanakhuha, Oliver Resch, Ralf Mikut, Veit Hagenmeyer
benchmarks
More like this →
Hugging Face Trending Papers
Jul 2

Probabilistic Low-Voltage Peak Load Forecasting with Time Series Foundation Models Evaluated on Application-Oriented Metrics

Low-voltage load forecasting is an important component in current and future energy systems with a high degree of electrification and decentralized generation. However, current forecasting methods require significant manual effort, often lack uncertainty estimation and proper peak prediction, and they are often not adequately evaluated in terms of grid requirements.

benchmarks
More like this →