Researchers develop breakthrough forecasting method for electricity prices. A new temporal hierarchy forecasting approach improves day-ahead electricity price predictions by up to thirteen percent across German and Spanish power markets. The technique, tested over four challenging years from 2021 to 2024, reconciles hourly and multi-hour block product forecasts simultaneously. The method works across multiple model types including neural networks, gradient-boosted trees, and transformers. With block product trading becoming increasingly common in energy markets, this approach offers significant accuracy gains at minimal computational cost, making it practical for daily trading operations and market participants seeking better price prediction capabilities.
Post from MarketNews_en
Log in to interact with content.