Forecasting Electricity Spot Market Merit-Order Curves with Functional Time Series Modeling

This study introduces a novel approach for forecasting merit-order curves in electricity spot markets by leveraging functional principal component analysis (FPCA) to efficiently represent a pair of supply and demand curves in a vector space and employing multivariate time series models for their prediction. Applied to the Italian day-ahead market during the 2023-2024 period, our approach generates accurate supply and demand curves forecast, and despite not being explicitly optimized for price forecasting, yields price forecasts which outperform state-of-the-art price-based models, highlighting the benefits of a curve-driven methodology.