A1237
Title: Forecasting electricity prices using bid data in times of distress
Authors: Ainhoa Zarraga - University of the Basque Country (Spain) [presenting]
Aitor Ciarreta - University of the Basque Country (Spain)
Blanca Martinez - Complutense University of Madrid (Spain)
Abstract: In the context of ongoing market reforms within the European Union, developing more accurate electricity price forecasting techniques is becoming increasingly essential for all market participants. The increasing penetration of renewables, coupled with the geopolitical tensions between exporters and importers of raw energy materials that emerged in 2021, has resulted in growing pressure on investors to adopt effective hedging strategies to protect their assets in this turbulent economic environment. The aim is to propose a price forecasting approach based on publicly available auction data to fit the supply and demand electricity curves of the Iberian electricity market for the period 2021-2024. First, fractional polynomial and logistic functions are fit to historical sales and purchase bidding data to estimate the equilibrium prices. Secondly, several time series models are specified and estimated for the historical and estimated prices. Thirdly, a rolling window is used to estimate models for both time series prices and forecast one-day-ahead prices for 2024. In-sample and out-of-sample error criteria are used. Results show that using fractional polynomials and logistic functions accurately replicates the observed prices. The out-of-sample forecasting analysis shows that the fractional polynomial functions outperform not only the naive model, but also the models based on historical prices.
