What candidates report
XTX publishes its openings rather than a stage-by-stage process, so this section is candidate reporting and is hedged as such.
Candidates commonly describe a short introductory call, an online test covering statistics, probability, general mathematics and computer science or algorithms, and then several interviews with quantitative researchers working through problems in the same areas. The reported process runs to four or five rounds.
The consistent theme is depth-testing rather than breadth. Candidates describe being asked to discuss recent machine-learning papers they have read, research directions they find interesting, and how they keep current - questions designed to distinguish genuine expertise from familiarity with the vocabulary.
Reported selectivity is high, and the online test is described as the sharpest filter in the process.
What to prepare
Statistics and probability with real precision. Reported to carry the technical weight of the early stages.
Algorithms and computer science. Named consistently alongside the mathematics.
A genuine reading position. If you are asked which recent papers interest you and why, a rehearsed list is worse than one paper you can actually discuss - what it claims, how it argues it, and what you think it gets wrong.
Depth over coverage. A process built to detect surface familiarity rewards knowing a smaller number of things properly.
Practise the same formats
The question sets below are ours, written in the formats this process is reported to use. They are not XTX Markets' questions, and nothing here reproduces any part of its assessment.