Technical Indicators for Hourly Energy Market Trading

Catherine McHugh, Sonya Coleman, Dermot Kerr

Research output: Contribution to conferencePaperpeer-review

135 Downloads (Pure)

Abstract

Financial trading often combines machine learning and technical indicators to accurately predict future market prices. Energy data and financial data have similar features; therefore, this research derives eight electricity price technical indicators to help control spending and reduce trading costs for the Integrated Single Electricity Market in Ireland. The proposed technical indicators were derived from electricity price data, collected on an hourly basis from February until November 2019, and used to train three regression machine learning algorithms (Random Forest, Gradient Boosting, and Extreme Gradient Boosting). The results for each of the regression algorithms were first compared using one model for all trading periods. The Random Forest algorithm was then trained with the same technical indicators for each of the 24 hours periods individually to see if an hourly approach enhanced model performance. The proposed technical indicators accurately predict electricity prices and overall accuracy was greatly improved using separate hourly forecasting models.
Original languageEnglish
Pages72-77
Number of pages6
Publication statusPublished (in print/issue) - 25 Oct 2020
EventDATA ANALYTICS 2020: The Ninth International Conference on Data Analytics - Nice, Nice, France
Duration: 25 Oct 202029 Oct 2020
Conference number: 9
https://www.iaria.org/conferences2020/DATAANALYTICS20.html

Conference

ConferenceDATA ANALYTICS 2020
Abbreviated titleIARIA
Country/TerritoryFrance
CityNice
Period25/10/2029/10/20
Internet address

Keywords

  • Hourly Forecasting
  • Machine Learning
  • Technical Indicators
  • Energy Market

Fingerprint

Dive into the research topics of 'Technical Indicators for Hourly Energy Market Trading'. Together they form a unique fingerprint.

Cite this