Winners of the Econometric Game 2025 Will Be the Guests of the QFRG and DSLab Seminar
Michał Künstler, Jakub Bandurski, Eliza Hałatek and Adam Łaziński – winners of the Econometric Game 2025, will be the special guests at the upcoming seminar organized by the QFRG and DSLab centres.
The seminar will focus on the topic „How to win world championships in econometrics”. During the event, the speakers will introduce the idea behind the competition, which has been organized for 25 years by the University of Amsterdam, and present their winning solution, entitled: „Is smooth Energiewende possible? Adjusting daily, weekly and monthly policies to minimize electricity congestion – the case of Germany”.
We warmly invite you to attend on April 28, 2025 at 18:30. The meeting will be held in a hybrid format: in person at the Faculty of Economic Sciences (Room B002) and online via the Zoom platform.
The presentation will last approximately 45 minutes and will be followed by an open discussion.
Link to the meeting: https://uw-edu-pl.zoom.us/j/97277803402
(Please arrive/log by 18:20)
The abstract of the presentation is available below.
---
Econometric Game is an annual international competition in the field of econometrics and data analysis organized in the University of Amsterdam since 1999. Due to its prestige, the event is often called the world championships in econometrics. Four person groups of master and PhD students invited from best universities around the world compete for three days to solve a real economic task. In 2025 the team from the Faculty of Economic Sciences, University of Warsaw (Michał Kunstler, Jakub Bandurski, Eliza Hałatek and Adam Łaziński) won the game competing against 30 other universities (including Harvard, Cambridge and Oxford). The case topic this year focused on predicting power grid congestion in the German TenneT DE electricity network as part of the broader Energiewende initiative. Our team applied a combination of machine learning and econometric techniques, including CatBoost, LASSO regression, and ensemble modeling, to forecast both up and down congestion from 2020 to 2023. The solution featured rigorous data analysis, interpretability through explainable AI, and policy recommendations across short-, medium-, and long-term horizons to support more efficient energy redispatch and integration of renewable sources. We will present the path that led us to victory, including data that we used, deep-dive into algorithms applied and XAI techniques employed.