Man AHL is the systematic investment engine of Man Group, one of the world's largest listed alternative asset managers, with roots going back to AHL's pioneering trend-following strategies in the late 1980s. Based in London and deeply integrated with the academic world through the Oxford-Man Institute, AHL runs a researcher interview that is distinctive in two ways candidates report: there is no speed-arithmetic gauntlet, and the centrepiece is a code review of your own take-home.
AHL explicitly does not run the market-maker playbook: no mental-math sprints, no market-making games. The process is built around an open-ended statistical take-home judged on design, then defended line by line.
- 1Recruiter screen and online tests. Aptitude and coding tests per the firm's graduate timeline.
- 2Take-home problem. An open-ended statistical or ML problem, reported at 24 to 48 hours.
- 3Code review call. A phone discussion that is a structured review of your own submission.
- 4Technical rounds. Coding, machine learning, and portfolio construction.
- 5Strategy round. Alpha sources and research discussion.
Stages 1 and 2: Tests and the take-home
The early filters are standard aptitude and coding tests. The take-home is where the evaluation really starts: an open-ended statistical or ML problem with room for design decisions, judged on how you frame and validate the work, not just whether it runs.
How to prepare
Write the take-home the way the firm will read it: clean Python with numpy, pandas and scikit-learn, a defensible validation scheme, and a short written account of what you checked. The validation course is the playbook, and all three problems below were built from the topics this firm's process reports: time series, and validation done properly on it:
Stage 3: The code review of your own work
The distinctive AHL round: interviewers walk through your submission and ask why, at every level. Why this model family, why this split, why this data handling, what breaks it. It is the project-defence format applied to work you did days earlier, which removes the one excuse every thesis defence allows: that it was a long time ago.
Prepare by reviewing your own submission adversarially before the call. Every shortcut you took is a question you should answer first, and volunteering a limitation with its fix reads far better than having it extracted.
Stages 4 and 5: Technical and strategy rounds
The reported technical topics: probability, time series, statistics, and clean Python. Time series deserves specific attention here; AHL's heritage is systematic trend-following, and the statistics interview reflects a firm that lives on temporal structure. The ML round covers the standard model families plus the methodology layer, and portfolio construction appears as its own subject: sizing, diversification, and how signals combine, which our signal-to-position lessons cover. The strategy round asks where you think alpha comes from; have a considered answer rather than a rehearsed one.
Key tips for success
- Budget real time for the take-home and spend a fixed slice of it on validation and write-up; that slice is what the code review grades.
- Drill time series properly: autocorrelation, stationarity, and why temporal data breaks textbook cross-validation.
- Do not import a market-maker prep plan. The process-type distinction matters more here than at almost any other firm.
Closing remarks
AHL's process is the cleanest example of interviewing by inspection: they watch you do research, then discuss the research you did. Everything reduces to one preparation: do the take-home well, know exactly why you did it that way, and be the first person in the room to point at its weakest part.