Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Day-Ahead Electricity Price Forecasting Using Merit-Order Curves Time Series

Guillaume Koechlin, Filippo Bovera, Piercesare Secchi

Abstract

We introduce a general, simple, and computationally efficient framework for predicting day-ahead supply and demand merit-order curves, from which both point and probabilistic electricity price forecasts can be derived. We conduct a rigorous empirical comparison of price forecasting performance between the proposed curve-based model, i.e., derived from predicted merit-order curves, and state-of-the-art price-based models that directly forecast the clearing price, using data from the Italian day-ahead market over the 2023-2024 period. Our results show that the proposed curve-based approach significantly improves both point and probabilistic price forecasting accuracy relative to price-based approaches, with average gains of approximately 5%, and improvements of up to 10% during mid-day hours, when prices occasionally drop due to high renewable generation and low demand.

Day-Ahead Electricity Price Forecasting Using Merit-Order Curves Time Series

Abstract

We introduce a general, simple, and computationally efficient framework for predicting day-ahead supply and demand merit-order curves, from which both point and probabilistic electricity price forecasts can be derived. We conduct a rigorous empirical comparison of price forecasting performance between the proposed curve-based model, i.e., derived from predicted merit-order curves, and state-of-the-art price-based models that directly forecast the clearing price, using data from the Italian day-ahead market over the 2023-2024 period. Our results show that the proposed curve-based approach significantly improves both point and probabilistic price forecasting accuracy relative to price-based approaches, with average gains of approximately 5%, and improvements of up to 10% during mid-day hours, when prices occasionally drop due to high renewable generation and low demand.

Paper Structure

This paper contains 23 sections, 18 equations, 15 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Context
  3. Knowledge gap
  4. Contributions
  5. Methodology
  6. Curves definition
  7. Curves representation
  8. Forecasting the curves' vector representation
  9. Clearing price point and probabilistic forecasting
  10. Benchmark curves and price forecasting models
  11. Forecasts evaluation
  12. Daily recalibration
  13. Results & Discussion
  14. Curves forecasting
  15. Price point forecasting
  16. ...and 8 more sections

Figures (15)

  • Figure 1: (Color optional) Example of a double uniform price electricity auction with supply and demand merit-order curves.
  • Figure 2: (Color optional) Example of ZST reconstruction for a supply-demand curves pair with $K_s=9$ dimensions for supply and $K_d=5$ for demand. The vector representation is given by the first-order differences of the $y$-coordinates of the red dots. The fixed price grid corresponds to their $x$-coordinates.
  • Figure 3: (Color optional) Performance of curves prediction measured with the squared correlation function $R^2(p)$
  • Figure 4: (Color optional) Results of the Diebold-Mariano test for the difference in price point forecasting performance performed (a) on average daily errors (b) at the hour level. The alternative hypothesis is that models on the $x$-axis outperform those on the $y$-axis (one-sided test).
  • Figure 5: (Color optional) Mean absolute error of the three best performing models within each category (price-based and curve-based with either ziel_electricity_2016 or FPCA representation) for each hour of the day.
  • ...and 10 more figures