What quant interview questions should I expect?

Every quant loop, whether for a trading desk, a research seat, or a risk team, opens with probability. The questions are old and still effective because they show how you think under a small amount of pressure. Expect coin and dice problems (the expected number of flips until a pattern appears, the probability of one player winning a race to a score), conditional probability puzzles that punish sloppy conditioning, expected-value questions framed as bets where you must also say whether you would take the bet, and combinatorics with a clean counting argument.

• State your approach first. Name the tool: linearity of expectation, symmetry, conditioning on the first step, a recursion. Interviewers listen for the choice of tool as much as the arithmetic.
• Brainteasers. Classic puzzles about weighing coins, crossing bridges, or ants on a stick reward recognising the trick; if you know the puzzle, say so and solve it anyway, then expect a variant.
• Estimation. Fermi-style questions where the reasoning chain matters and the final number is checked only for order of magnitude.
• Mental arithmetic. Trading interviews run timed arithmetic and sequence tests; speed comes from practice, not talent.

A useful discipline: talk through the setup, commit to an approach, compute, then sanity-check the answer against a limiting case. Candidates who check their own answer before the interviewer does are remembered.

You are comfortable with the puzzles and worried the second round will go into mathematics you last saw in a course. It will, and the depth depends on the seat, so this section separates what every role asks from what research and pricing roles add.

• Statistics for every role. Bias and variance, the law of large numbers versus the central limit theorem, hypothesis testing and what a p-value is not, correlation versus dependence, and how to spot overfitting in a backtest.
• Linear regression. The assumptions behind ordinary least squares, what happens with correlated regressors, why you might regularise, and how to read residuals. Research seats ask you to derive the estimator or explain it geometrically.
• Time series. Stationarity, autocorrelation, and why a strategy that looks good in-sample often fails out of sample.
• Stochastic calculus for pricing roles. Brownian motion and its properties, martingales, Ito's lemma, and the intuition behind risk-neutral pricing and the Black-Scholes assumptions. Expect to explain what the assumptions leave out rather than to reproduce a full derivation.
• Options intuition. What delta, gamma, and vega mean for a position, and how a hedge behaves as the underlying moves.

Answer with the idea and its failure mode. Explaining why a p-value below a threshold does not make a strategy profitable is worth more than reciting the definition.

Coding rounds are practical rather than framework-heavy: implement a data structure, process a stream of prices, compute a rolling statistic efficiently, or write a small simulation. Python is standard; C++ appears for latency-sensitive desks, where questions about memory layout and undefined behaviour follow. Interviewers care about correctness, clarity, and complexity, and they notice whether you test edge cases without being asked.

Trading desks add market-making games. A representative one: a candidate interviewing for a junior trader seat at a proprietary trading firm is asked to quote a bid and an ask on the number of windows in a large office building, then the interviewer trades against the quote, adds information, and asks for a new market. The strong candidate widens or tightens the spread with the uncertainty, updates on the interviewer's trades rather than ignoring them, and keeps track of the position and its risk. The game measures calibration and composure, not trivia about buildings.

Rehearsing these aloud is the only preparation that transfers; the mock interview mode runs questions with follow-ups, and the other role and topic banks sit under interview questions by role and topic.

Only for the conversational parts, and the limits matter more here than in most interviews. SubcueAI's native macOS and Windows desktop app captures system audio and your microphone and shows short suggestions in a local overlay, so when an interviewer asks what Ito's lemma is for or what a scoped hedge does as gamma changes, the idea is on your screen while you explain it in your own words. The browser extension covers browser-tab calls on Chrome and Edge by capturing the meeting tab's audio only. No bot joins the call and nothing is injected into the meeting page; setup is on the tutorial page.

What it cannot do: timed online arithmetic and coding tests, proctored assessments, recorded screens, company-managed laptops, and any round where you must compute quickly on a whiteboard or think aloud through a puzzle. Those rounds are the interview, and they are your own work. The assistant fits the terminology and intuition questions above; the honest boundaries are on the detectability cluster.