Wednesday, February 10, 2016

EEM2016 Electricity Price Forecasting Competition

Electricity price forecasting is a very difficult problem. Although electricity demand is a major driving factor of the prices most of the time, the price spikes, which happen only a few hours in a year, are typically the result of many factors, such as transmission congestion, generation outages and behaviors of market participants. Moreover, these spiky periods may put the power companies into higher risk than the risk of off-peak periods combined.

Most academic studies are focusing on statistical and artificial intelligence techniques, which can hardly predict the spikes accurately in practice. On the other hand, simulation methods that involves modeling of the market and power flows usually require very specialized skills and software packages, which can be too expensive to implement to some organizations. Check out my Foresight article (Energy Forecasting: Past, Present and Future) for a bit more discussion on electricity price forecasting, or Rafal Weron's review paper on price forecasting for an even more comprehensive discussion.

When organizing GEFCom2014, we were hoping to launch a realistic price forecasting track. Due to various constraints, we finally decided to launch a simplified version - forecasting electricity prices based on historical prices and load, and load forecasts. Nevertheless, organizing a "real" price forecasting has been on my wishlist since then.

This afternoon I received an email from Claudio Monteiro introducing a price forecasting competition they are organizing as part of the the 13th European Energy Market Conference (EEM2016) activities. They are taking the price forecasting competition to the next level, by making it realistic and similar to the real-world settings. Anyway, I'm so delighted and excited to spread the message here:
The EEM2016 Organizing Committee in collaboration with SmartWatt are launching a price forecast competition. The objective is to provide a dynamic and realistic gaming context, to involve the scientific, academic and industry community in order to address a real price forecast problem.
The challenge will be to forecast the hourly spot price of the Iberian Electricity Market, MIBEL, on a daily rolling basis, for the 24 hours of the 5 days ahead. Over 14 days of competition, from 4th to 17th of April 2016, until 10.00 UTC, competitors must submit the forecast for the 5 days ahead. Historical hourly data will be provided for 2015.
The winner will receive 1000 € and the 3 highest ranked forecasters will have free access to the EEM2016 Conference. Complete information about this competition can be obtained HERE (PDF). There is also a facebook page related with the competition.
SmartWatt is developing a web benchmark platform to manage this specific forecast competition. The registration in Complatt Price Forecasting Competition platform will be opened on the 15th of February. The full functionality will be launched on March 15th. 

Happy electricity price forecasting! 

1 comment:

Note that you may link to your LinkedIn profile if you choose Name/URL option.