US Midterms Tracker
A daily updated overview of the 2026 congressional races.
Data journalist
My name is Marcel Pauly, and I’m a data journalist based in Hamburg, Germany, with more than a decade of experience. Since 2017, I’ve worked at DER SPIEGEL, a leading German news magazine, where I’m part of the Data and Visualization desk.
My beat is political data: I investigate transparency and influence in German politics, analyze election results, and model polling averages. I code in Python, scrape websites, and use LLMs to turn unstructured text and documents into structured datasets. I build automated dashboards and am responsible for the technology that updates election results on SPIEGEL.de in real time on election nights.
After studying communication studies, political science, and German studies at Johannes Gutenberg University Mainz, I trained at the Henri Nannen School of Journalism in Hamburg. I honed my data journalism skills through the Lede Program at Columbia Journalism School in New York City.
I shared responsibility for the data reporting on investigations that have won awards including the Georg von Holtzbrinck Prize for Business Journalism and the Helmut Schmidt Journalism Prize. Other projects were nominated for the German Reporter Award and the Grimme Online Award.
A selection of investigations and projects. The articles are in German.
A daily updated overview of the 2026 congressional races.
Poll trend and nowcast for the German federal election, modeled in Stan.
Ahead of elections, we simulate outcomes based on polls and calculate the probabilities of the possible coalitions.
We scrape the free-text disclosures of Bundestag members, parse them with the help of an LLM and analyze them.
I analyzed which voter groups carried Friedrich Merz into the Chancellery.
I built the data pipeline that delivers projections and results to SPIEGEL.de within fractions of a second on election nights.
We digitized the domestic intelligence agency’s assessment of the AfD and made it searchable.
We used AI to digitize the Nazi Party membership card index and made it searchable.
We processed 154 years of population data – for every single municipality in Germany.
Every half hour, we process weather data from hundreds of measuring stations and put it into historical context.
Schufa data from more than 2,000 consumers reveals how the algorithm of Germany’s most important credit bureau works.
Adult education centers offer questionable courses. We scraped their programs and analyzed them.
How a young mathematician developed an algorithm to find the perfect constituency boundaries.
We analyzed TikTok’s recommendation algorithm and uncovered how it works.
Employees of the Federal Statistical Office make serious allegations against the agency’s head. I spoke with several of them.
Email: marcel.pauly@spiegel.de