🤖 AI Summary
This paper addresses the limited interpretability of data-driven models and the insufficient fidelity of fundamental models in electricity price forecasting. We propose a novel data-driven marginal merit-order model that integrates structural economic principles with machine learning. Structurally grounded in classical merit-order theory—here innovatively embedded as a learnable special case—the model infers key parameters—including plant efficiency, bidding strategies, and available capacity—from historical market data via inverse modeling. It further introduces three original modules: hydroelectric scheduling, cross-border power flow modeling, and capacity underreporting correction. The resulting framework achieves both high predictive accuracy and strong economic interpretability. In empirical evaluation on the German day-ahead market, it significantly outperforms conventional fundamental models and state-of-the-art machine learning approaches. The model enables interpretable marginal technology identification, fuel-switching analysis, and unit-level generation attribution—bridging the gap between black-box forecasting and actionable market insights.
📝 Abstract
Electricity price forecasting approaches generally fall into two categories: data-driven models, which learn from historical patterns, or fundamental models, which simulate market mechanisms. We propose a novel and highly efficient data-driven merit order model that integrates both paradigms. The model embeds the classical expert-based merit order as a nested special case, allowing all key parameters, such as plant efficiencies, bidding behavior, and available capacities, to be estimated directly from historical data, rather than assumed. We further enhance the model with critical embedded extensions such as hydro power, cross-border flows and corrections for underreported capacities, which considerably improve forecasting accuracy. Applied to the German day-ahead market, our model outperforms both classic fundamental and state-of-the-art machine learning models. It retains the interpretability of fundamental models, offering insights into marginal technologies, fuel switches, and dispatch patterns, elements which are typically inaccessible to black-box machine learning approaches. This transparency and high computational efficiency make it a promising new direction for electricity price modeling.