ML Quantitative Researcher

Design the predictive signals behind live systematic equity strategies, where the research you run this week is trading the next.

Our client is a quant asset manager based in Germany, trading equity strategies alongside asset-allocation books across developed markets. You'd join a research pod that trains machine-learning models to rank stocks by expected relative return, and the signals feeding those models would be your patch to own.

This is not internal tooling or analytics on the side. The signals you build are the trade. And the team is just as interested in the economic reason a signal works as in the backtest that proves it does, so the role sits right where research, market intuition, and engineering meet. You'd begin with single signals end-to-end, and take on more as you find your feet.

Your role:

• Search datasets and investment universes for signals with genuine out-of-sample staying power, weighing them on both statistical strength and economic logic
• Convert messy source data (prices, volumes, fundamentals, text) into dependable signals, packaged as compact, unit-tested, configurable components in the research pipeline, with point-in-time discipline throughout
• Blend clusters of correlated signals into a handful of resilient composites per universe, weighting for originality and information, and stripping out drift that masquerades as stock selection
• Put the research cycle on rails: build agentic/LLM workflows that propose ideas, sweep parameter grids, and produce their own evaluation write-ups

Your profile:

• A master's or doctorate in a quantitative discipline: maths, physics, computer science, financial engineering, statistics, or similar
• Real command of statistics plus some econometrics, and a level head when the data is noisy and real
• Comfort shipping Python in a team setting: version control, reviews, and clean, thoroughly tested code (the stack leans on Pydantic and pytest)
• Practical fluency in a modern dataframe library, Polars or pandas

Interested? Please click the apply button.