Forecasting the Future: Volue Replaces Its Spot Model To Meet the Challenges of the New Energy Market
Every 15 minutes, Volue publishes one of its short-term energy forecasts. The spot forecast is used by power professionals, traders, investors, and analysts across Europe.
“We provide a comprehensive analysis close to real-time,” says Marianne Aanonsen, Head of Quantitative Analysis at Insight by Volue.
But the energy market is changing and Aanonsen’s team is replacing the spot model, the engine behind the spot forecast, to meet the challenges of the new power market.
“Europe is adding renewables to its energy system on a massive scale, interconnecting power grids, and integrating more countries into a single power market. As the energy industry transforms, so does our spot model.”
This week, her team replaced the legacy InstantSpot model with the new model, SpotEx.
“We are using the same curve IDs which means that there will be no formal change on the customer side. However, over time, our customers will see solid results. It’s like upgrading the engine of your car. It’s all about quality and preparing for the future.”
Ready for the extension of the Flow-Based Market Coupling
One big challenge which SpotEx will answer is the upcoming Core Flow-Based Market Coupling – the integration of more European countries into one market.
“In Spring 2022, seven more countries are joining the Flow-Based Market Coupling algorithm. There is a plan to add Flow-Based Market Coupling for the Nordics, too.“
In a coupled market, demand and supply orders are not confined to a single country. Energy transactions can involve sellers and buyers from different areas. Market Coupling improves market liquidity and makes electricity prices less volatile.
“Flow-Based Market Coupling increases complexity big time. It is data-intensive and requires more power to calculate and forecast than individual, single-price zones. This is what we need to handle. SpotEx is ready for this.”
Improve and stabilise the model for the future
The project comes in steps.
In addition to replacing the spot model, the team has done a lot of work to improve the fundamentals data input.
“Already in phase one, we have better quality results than before. SpotEx is outperforming InstantSpot. Now, we are preparing for the second phase.”
Aanonsen says that in the new reality, manual processes don’t work anymore because so much is happening close to real-time.
Her team has worked with machine learning and AI for their various models for at least three years. Now they are bringing machine learning to SpotEx.
“Basically, SpotEx says goodbye to linear exchange assumptions and responses. It allows non-linear price sensitivity in the exchange algorithm. It uses machine learning algorithms to make effective merit-order curves.”
SpotEx is more dynamic and performs better during periods with high volatility capturing the price more accurately. It reflects better the uncertainty in the individual weather scenarios of the ensemble forecasts.
It uses an approach very similar to the official Euphemia implementation, the algorithm used to calculate energy allocation and electricity prices across Europe.
The new model also gives much more flexibility to forecast different countries separately.
“Then comes the next part, going through all the different countries and making all of them better one by one. We are also adding more countries.”
Aanonsen expects that customers won’t see the change day-by-day but over time the quality will be better.
“It’s different for different time horizons. For the day-ahead forecast, it’s marginally better. But further ahead it’s much better. Three days ahead, five days ahead, that’s where you really see the difference.”
Meanwhile, the work on SpotEx continues.
“This is a big project that has required collaboration across the organisation. Jan-Philipp Eisenbach has contributed the most to the development, along with Tim Becker, Till Ilic, Andreas Malmgård, and Gisle Tveit. They are supported by market experts Odd Gunnar Jakobsen, Silvia Messa, Katinka Bogaard and Anamaria Toebe. In the next phase, the knowledge of our market experts will be even more important.”