DG 0199/DIA/2019/48

Forecasting wholesale electricity prices for risk management purposes: Using regularization and forecast averaging

Grant no.: 0199/DIA/2019/48
Funding agency: Ministry of Science and Higher Education, Poland
Funding scheme: Diamond Grant
Funding period: 2.10.2019-1.10.2023 (48 months)
Budget: 180 000 PLN
Title in Polish: Prognozowanie hurtowych cen energii elektrycznej na potrzeby zarządzania ryzykiem: Wykorzystanie regularyzacji oraz uśredniania prognoz

Research team:

Principal Investigator (Kierownik):

Research Supervisor (Opiekun naukowy):

Collaborators (Współpracownicy):

– Ph.D. / M.Sc. / B.Sc. student

Aims and scope:

The project aims to develop methods and tools in three key areas of electricity price forecasting. First, it focuses on probabilistic forecasting to predict not only point estimates but also intervals and distributions, aiding in risk management and trading. Second, it explores the selection of fundamental input variables like demand, weather, and fuel costs, crucial for improving model accuracy. Lastly, it emphasizes forecast averaging across different models, reducing risk and enhancing predictive performance. These efforts will contribute to both theoretical advancements and practical improvements in portfolio and risk management in the energy sector. The project is based on the concept of regularization, which has been widely developed but has rarely been used in electricity price forecasting. Existing studies have focused on point forecasts, but this project introduces a novel application of regularization. To achieve its goals, the project adopts an integrated approach with three interrelated tasks running in parallel:

Tasks:

  1. Developing algorithms for probabilistic electricity price forecasting, utilizing techniques such as quantile regression and regularization.
  2. Assessing the impact of penalty functions in regularization methods on forecast accuracy and optimize the selection of input variables.
  3. Developing algorithms for averaging forecasts from individual models, incorporating regularization methods.

Publications:

Peer-reviewed articles in JCR-listed journals

2024 (1), 2023 (1), 2022 (0), 2021 (1), 2020 (2)

  1. B. Uniejewski (2024) Regularization for electricity price forecasting, Operations Research and Decisions 34, 3 (doi: 10.37190/ord240314).
  2. B. Uniejewski, K. Maciejowska (2023) LASSO principal component averaging: a fully automated approach for point forecast pooling, International Journal of Forecasting 39, 4 (doi: 10.1016/j.ijforecast.2022.09.004).
  3. B. Uniejewski, R. Weron (2021) Regularized quantile regression averaging for probabilistic electricity price forecasting, Energy Economics 95, 105121 (doi: 10.1016/j.eneco.2021.105121).
  4. K. Maciejowska, B. Uniejewski, T. Serafin (2020) PCA forecast averaging – predicting day-ahead and intraday electricity prices, Energies 13, 14 (doi: 10.3390/en13143530).
  5. G. Marcjasz, B. Uniejewski, R. Weron (2020) Beating the naive – combining LASSO with naïve intraday electricity price forecasts., Energies 13, 7 (doi: 10.3390/en13071667).

Conference papers

2024 (0), 2023 (1), 2020-22 (0)

  1. B. Uniejewski (2023) Enhancing accuracy of probabilistic electricity price forecasting: A comparative study of novel Quantile Regression Averaging generalization, 19th International Conference on the European Energy Market (EEM23), (doi: 10.1109/EEM58374.2023.10161748).