A new paper by Florian Ziel and Paul Ghelasi introduces a new electricity price model that combines energy economic theory with data‑driven methods in a transparent and interpretable way.
The new study "A data-driven merit order: Learning a fundamental electricity price model" in the journal Energy Economics examines how electricity prices are formed and how they can be forecasted accurately in today’s power markets.

Electricity price formation
In electricity markets, prices are determined in the same way as in basic economics: at the intersection of supply and demand. Supply is provided by power plants (such as wind, solar, coal, and gas), while demand corresponds to electricity load. Power plants are ranked from lowest to highest production costs—the merit order—and the most expensive plant required to meet demand sets the market price.

Modeling approach
Building on this principle, the proposed model is specifically designed for electricity prices. A simplified, highly efficient country-wide merit order is constructed and then calibrated to historical prices. Unlike econometric or machine-learning models, it not only remains fully transparent and interpretable, but also provide causal conclusions and allows key power‑plant and bidding parameters to be estimated directly from market data.

Results
Applied to the German day‑ahead electricity market, the model achieves forecasting accuracy comparable to state‑of‑the‑art machine‑learning approaches, despite not using autoregressive price effects.

Insights and outlook
The model also provides structural insights into price‑setting technologies, fuel‑switching behavior, and dispatch patterns. As a fundamental model, it can be extended to long‑term and scenario‑based analyses, probabilistic forecasting, and the modeling of interconnected electricity markets.

Further information
Further information can be found under Publications.
Click here to access the publication. https://doi.org/10.1016/j.eneco.2025.109114