22 November 2024
Predico: collaborative forecasting platform
Information on this page is a subject to further studies and analysis
A platform designed to enhance forecasting accuracy and efficiency in the energy industry by introducing a multilateral, collaborative market for renewable energy forecasts.
What is Predico?
Predico is a collaborative forecasting market platform enabling entities with common interests to procure and sell forecasts. The platform was developed to lower overhead costs for all parties involved by leveraging market mechanisms and ensemble forecasting to improve forecasting accuracy and properly align forecaster incentives.
Why participate as a forecast buyer?
Accuracy: Predico is more accurate than standard bilateral contracting for three main reasons:
-
Ensembles outperform individual forecasts because they:
-
Markets outperform bilateral contracts because they:
-
Simplicity & Efficiency:
Why participate as a forecast provider?
-
Remuneration
-
Simplicity & Efficiency
-
Market Access
-
Zero SLAs
How Predico works and how to get involved
-
How does Predico work?
-
How do I become a participant?
Project timeline
- November 2024 - March 2025: Test phase by running a proof of concept for the Belgian offshore wind production
- March 2025: Opening the platform to external buyers and markets.
Participate in the Belgian Offshore Wind Generation Market
Our first market is for Belgian offshore wind energy generation.
- Go-live: November 2024.
- Market value: €7k/month for total remuneration. This amount would be spread between forecasters participating in the market based on their accuracy rank and information contribution. Realistically, the best forecaster on the market could earn about €3k/month from the forecast. More details can be found in the documentation.
- Target variable (also available on Predico): The ‘measured and upscaled’ value is published at the following address.
- Resolution: 15 mins
- Horizon: Forecast at 10:00 European Central Time on D-1 for day D
- The target variable: The sum of all generation coming from the Belgian offshore wind area, not from a single wind farm.
- Forecast type: Probabilistic: P10, P50, P90.
Tips for forecasting this value:
- The historical values are available on Predico or Elia's Open Data Portal here.
- The near-real time values are available on Elia's Open Data Portal here.
- Wind turbines are unavailable sometimes. The schedules can be found here and here.
- We are the TSO and do not own or operate the turbines, only the lines. Therefore we don’t have the data about the wind turbines themselves. But you can start here to have a look at the wind farms in question.
- Note that on 22/05/2024 the data format for the above links changed: older data are also available on the Elia Open Data portal.
- Wind generation seems less correlated with system imbalance than solar. This means that the wind turbine operators don't react to imbalance markets as much (although the BRPs might have wind and solar in their portfolios and co-optimize). But in general this is a good thing for wind generation forecasting. But do be aware of the balancing market and how that affects operations.
- Bear in mind that above a certain wind speed, the turbines cut out as a safety measure. How this is done depends on the wind farm operator but is important for wind generation.
- Wind gusts or gustiness are often useful to include as a feature.
How to register
We’re looking for forecasts which are equal or superior to our currently published forecast (available at the data link above).
Rules for participation can be found in the Terms and Conditions – please read them before contacting us.
To register, send an email to predico@elia.be, providing the following details:
If you wish to participate in the performance-based compensation aspect of the market:
- Link to your forecasting company website.
- Client references
If you do not wish to participate in the performance-based compensation aspect of the market, but you do wish to submit forecasts, please confirm in writing in the email.
Platform connection details: Predico REST API (inesctec.pt).