A new benchmark called EnergyBench has been developed to evaluate energy forecasting models in real-time for European power markets. Unlike traditional methods that rely on backtests and fixed datasets, EnergyBench runs forecasts live daily, providing a more accurate comparison of model performance. The benchmark incorporates numerical weather prediction models and ranks models based on their error reduction compared to a baseline, with results indicating that foundation models can match or outperform trained models when used zero-shot with weather covariates. While model size has a limited impact, ensembles of covariate models perform best, though grid operators' forecasts still lead in wind and solar generation prediction. AI
IMPACT Provides a more robust evaluation framework for AI models in energy forecasting, potentially accelerating adoption and improving accuracy in power markets.
RANK_REASON Academic paper detailing a new benchmark for energy forecasting. [lever_c_demoted from research: ic=1 ai=0.7]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
