9 June 2022, 08:23
The improvement in forecasting capabilities is 18% compared to the previous model
Saluzzo (CN), 9 June 2022 – eVISO S.p.A. (EVS.MI) – an EGM-listed digital company with a proprietary artificial intelligence infrastructure that creates value for recurring consumers and producers of raw materials – announces the release of its new algorithm for forecasting the PUN (Single National Price) in the energy sector, which will enable the company to improve its ability to calculate exposures and guarantees.
The new algorithm, developed by the in-house “Algo Intelligence eVISO” department, allows for highly accurate analyses with a MAPE (Mean Absolute Average Error) margin of error improved by more than 18% compared to previous models – over the 12-month timeframe of 2021.
In particular, the forecast improved by more than 35% in December 2021, a month affected by the exceptional rise in the PUN.
In economic terms, during the calendar year 2021, the PUN rose dramatically, reaching a monthly average of €281.24/MWh in December, with a record high of €533.19/MWh compared to the annual average of €125.46/MWh in 2021. It was precisely in December 2021, a month characterised by high volatility, that the new model proved all its stability and robustness: indeed, the aforementioned 35% forecast improvement over the model used to date would have corresponded, in absolute terms, to 18.5€/MWh. The impact of the algorithm’s performance is all the more relevant in view of the expected price tension in the coming years, which are expected to be around 209€/MWh (price of the 2023 EEX baseload future) compared to an average of 52.32€/MWh for a pre-covid year such as 2019.
In summary, the PUN forecast provided by the new algorithm is useful for three reasons
- customers, both direct and resellers, are guaranteed greater stability and visibility in these periods of high volatility;
- the forecast data is used to calculate the exposures and guarantees that need to be provided to the market and this ensures greater financial stability and optimised use of resources;
- finally, eVISO has a reliable forecast to operate in the market even during disrupted data flow from the market (“sailing in the dark” scenario).
Carlo Cigna, head of eVISO’s Algo Intelligence department, commented: “The path to creating the new algorithm required several months of research by the Algo Intelligence eVISO team, which implemented the forecasting model using a new artificial intelligence methodology capable of capturing additional signals. Starting from the assumption that a curve is composed of many factors, but can be broken down in the sum of simpler factors, they worked on the breaking down of the curve that determines the value of the National PUN. To simplify, just as within a musical composition there are numerous sounds that can be traced back to different scores, we worked on finding the contributions of the individual instruments in the orchestra.”
Gianfranco Sorasio, CEO of eVISO, commented: “Artificial intelligence plays a key role in the energy transition because it allows us to predict market fluctuations in both price and demand more accurately thanks to a massive analysis of data acquired in real time. Over the past six months, the energy sector has seen repeated black swans. The improved forecasting algorithm has used the energy sector as a benchmark for subsequent applications on other commodities.”
***
This press release is available in the Investor Relations section of the website www.eviso.ai. Please also note that, for the dissemination of regulated information, the Company uses the eMarket SDIR circuit managed by Spafid Connect S.p.A., with registered office at Foro Buonaparte 10, Milan.