The Quadrature Capital Interview, Explained

Quadrature Capital is a London systematic fund with a deliberately low profile, top-of-market compensation, and — per candidate reports — an interview loop that resembles a research lab's hiring more than a trading floor's. (Candidate-reported; treat as orientation.)

The reported shape

• Technical screens on probability, statistics, and algorithms — leaning research-flavored: estimator reasoning, data-pitfall questions, and clean coding rather than brainteaser theater.
• ML and statistics depth for research roles: overfitting and validation under non-stationarity, regularization intuition, feature-selection reasoning — the ML question family, asked seriously.
• Serious software engineering — Quadrature hires engineer-researchers; expect real code quality expectations, not pseudocode tolerance, in the same register as LeetCode-medium/hard with engineering judgment.
• Research discussion / take-home style stages reported for some tracks — walking through how you would investigate a signal or debug a model, evaluated on process.

What the loop selects for

The consistent theme in reports: epistemic care. Questions push on how you would know a result is real — validation design, leakage, multiple-testing awareness ("if you test 100 signals at 5%, what do you expect?"), regime change. Candidates who volunteer the failure modes of their own proposed approach reportedly do well; confident overclaiming does not.

Prep plan

• Statistics with teeth: the statistics bank and regression bank — estimators, hypothesis testing, and regression pathologies are the loop's center.
• ML reasoning: the ML bank, focusing on validation and overfitting narratives, not architecture trivia.
• Probability foundation: the probability bank for the screen layer.
• Coding at engineering standard: LeetCode mediums with attention to code cleanliness, plus the quant coding bank for numerically-flavored problems.