The company has developed an algorithmic model that integrates a vast amount of unstructured and textual information to improve the forecasting of energy market price fluctuations.
Jean-Michel Gaudron
[This article is part of a content series developed in collaboration with FEDIL, showcasing how artificial intelligence is contributing to the digital transformation of Luxembourg’s economy.]
Founded in Luxembourg in 2022, E.E.M.M. (European Energy Market Makers) specialises in providing advanced trading strategies tailored to all stakeholders in the energy industry, including producers, traders, energy communities, cooperatives, grid operators and financial institutions. The company harnesses sophisticated algorithms powered by artificial intelligence to build models that forecast renewable energy production (wind and photovoltaic) as well as energy market prices.
“Our models analyse and process millions of data points, including historical energy production, weather forecasts and market price trends,” says Klemens Klotzner, the company’s Co-Founder and CEO.
More than 100 languages analysedAs powerful as it was, the existing model remained highly sensitive because it was strongly influenced by geopolitical risks, a major source of uncertainty that can directly affect energy prices and force traders to revise their strategies, sometimes at very short notice.
“Another challenge was the accurate forecasting of renewable energy production, since better predictions reduce balancing fees and benefit consumers. This required the incorporation of a vast amount of unstructured and textual information and therefore the development of a dedicated AI model specifically adapted to this use case.”
We significantly outperform conventional models, Prof. Dr Hans-Jörg von Mettenheim E.E.M.M.
To enhance market price prediction, the company implemented an AI-driven algorithm capable of processing news articles in more than 100 languages. This enables the system to detect geopolitical events within seconds and immediately adjust trading strategies. Given the size and data intensity of these models, E.E.M.M. relies on the computing power of MeluXina, Luxembourg’s national supercomputer.
“The key advantage, and our competitive edge, lies in the uniqueness of our approach,” explains Prof. Dr Hans-Jörg von Mettenheim, Scientific Advisor to E.E.M.M. “Unlike standard machine learning, which optimises numerical datasets quickly but with limited depth, our proprietary algorithms are designed to train advanced text-based models. This allows us to capture and process complex, unstructured information and significantly outperform conventional models.”
Better risk, better investmentIn practical terms, implementing this model has significantly increased the Sharpe Ratio, a measure that evaluates an investment’s performance while taking account of the level of risk involved.
As a result, the strategies developed deliver improved risk-adjusted returns and stronger overall investment performance.
E.E.M.M. is now focusing on expanding its solutions within the financial sector and building new partnerships. “We were recently awarded a contract by one of the largest green energy communities in Italy to implement our software,” notes Mr Klotzner.
Better predictions reduce balancing fees and benefit consumers. Klemens Klotzner, E.E.M.M.
“We also aim to adapt our solutions for smart city applications in order to attract and support a broader range of government clients. We are deeply committed to research, collaborating with leading scholars from institutions such as the Oxford Man Institute at the University of Oxford, and presenting our work at major international conferences and symposia.”
