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

Name: Evaluating Time Series Foundation Models for Electricity Price Forecasting: Contamination Risk, Distributional Shifts, and Covariate Dependence
Uploaded: 2026-10-08T12:19:19.667533+00:00
Duration: 434 s
Description: This study benchmarks time series foundation models (TSFMs) for electricity price forecasting using two datasets to reduce contamination risk. Researchers tested four TSFMs against classical and deep learning baselines, finding that TSFMs perform well but require covariate support. The best results came from simple ensembles combining the top TSFM and a domain-specific model.

This study benchmarks time series foundation models (TSFMs) for electricity price forecasting using two datasets to reduce contamination risk. Researchers tested four TSFMs against classical and deep learning baselines, finding that TSFMs perform well but require covariate support. The best results came from simple ensembles combining the top TSFM and a domain-specific model.

Open the full episode