Point72 / Cubist Interview Questions

Inside the Point72 / Cubist interview

Point72 is Steve Cohen's multi-strategy hedge fund, and Cubist is its systematic arm. Its Quant Researcher and Data Scientist loops emphasize statistical reasoning you can defend out loud: probability, regression and estimation, a working grasp of ML regularization, and clean coding.

What they test

The core is probability and expectation — Markov-chain hitting times, optimal stopping, and quick fair-value pricing of simple bets. Around it sits a distinctive statistics and regression block (variance computations, mean/median/mode ordering, slope algebra, collinearity) plus ML intuition on bagging and LASSO. A real coding set rounds it out, and brain-teasers screen for structured thinking.

The recurring shapes

Many problems reduce to a few moves: set up a recurrence or first-step decomposition for an expected count, exploit symmetry to collapse cases, or reason about an estimator's bias and variance. The regression questions reward knowing that a slope is just a scaled correlation, and the ML questions reward knowing what regularization does to collinear or redundant predictors.

How to approach

Narrate the model before computing: name the distribution, the conditioning, or the recurrence, then carry it to a clean closed form. For the systematic side, be ready to tie a statistical answer back to signals and data — why an estimator is consistent, when a series is stationary, what LASSO keeps. For coding, write correct, readable solutions and state the complexity.

Thirty problems leaning medium, with a spread of easy warm-ups and a handful of hard regression and coding questions to separate candidates.