Quant Trader vs Quant Researcher: What the Interviews Actually Test

Last updated: November 2026.

Quick answer: Quant trader interviews test speed and composure: timed mental math, fast probability, live market-making games and bet sizing. Quant researcher interviews test depth: statistics, regression pitfalls, a grilling on past research and moderate coding. Both share a probability core. Past that, prep for the loop your target firms actually run.

A lot of quant prep material puts both roles under one heading, "quant interviews," and hands you the same list of brainteasers. That costs you time. The two loops are filtering for different people, and a strong trader candidate and a strong researcher candidate can each fail the other's interview.

If you want the day-to-day view of the researcher seat first, read what quant researchers actually do all day. This post is about the interviews themselves: what each round measures, and where your prep hours should go.

How Is the Quant Trader Interview Different From the Quant Researcher Interview?

The cleanest public summary of the split is QuantVault's comparison of the two roles (updated July 2, 2026). In their words:

• Trader loops: "mental math screens, fast probability, market-making games, sizing questions"
• Researcher loops: "statistics/ML depth, regression pathologies, research discussions, moderate coding"

The same page sums up the selection logic in one line: "QT selects hard on speed (mental math, live games, composure), QR on depth."

Here is that split laid out round by round:

The last row is the one to remember. Trader interviewers are watching for someone who stalls under pressure. Researcher interviewers are watching for someone who gets fooled by their own data.

What Does a Quant Trader Interview Actually Test?

Every round puts a clock on you, and most of them make you commit to a number while someone pushes back. Expect these rounds:

• A timed arithmetic screen. TraderMath's practice page (checked November 2026) describes one widely discussed format as "80 mental-arithmetic questions in 8 minutes, with negative marking." Six seconds a question with a penalty for misses means guessing is a bad strategy. A steady pace with near-perfect accuracy beats a fast pace with errors.
• Fast probability and expected value. Short questions where the answer feeds a decision: would you take this bet, and at what price?
• Market-making games. You quote a two-sided price on something uncertain. The interviewer buys or sells against you, then adds information.
• Sizing questions. You know your edge on a bet. How much do you put on?

Sizing is where candidates who studied only puzzles get caught. Take a made-up example: a bet pays even money and you win 60% of the time. The textbook answer from the Kelly rule is to stake 20% of your bankroll, because at even money the fraction is your win probability minus your loss probability. A strong candidate gives that number and then says why they would bet less in practice: their 60% is an estimate, and overbetting an edge you have misjudged does far more damage than underbetting it.

How do you play the market-making game well?

Here is an original walkthrough on a clearly hypothetical quantity, the sum of two fair dice.

The fair value is 7, so you open with a market around it, say 6.5 bid and 7.5 ask. The interviewer buys at your ask. Now you ask yourself why they bought. Maybe they know something. You move your quote up a little, perhaps 7 bid, 8 ask. Then the interviewer tells you one die shows a 5. The fair value becomes 5 plus 3.5, which is 8.5, so your new market goes around 8.5, and the spread can tighten because half the variance is gone.

Compare two ways of handling that information step:

Weak:

"Okay, one die is a 5... I'll stay at 7 bid, 8 ask, since I already moved once."

Strong:

"With one die known at 5, fair value is 8.5. I'll go 8 bid, 9 ask. The remaining die still has spread, but half the variance is gone, so I can quote tighter than I did at the start."

The weak answer protects the last quote. The strong one throws it away the moment the facts change, which is exactly the habit the game is built to test.

What Does a Quant Researcher Interview Actually Test?

These rounds give you more time per question and then spend it probing whether you would catch your own mistake. Expect these rounds:

• Statistics and machine learning questions. Inference, regression, model validation, and what to do when you have run many tests at once.
• Regression pathologies. A question that looks routine until you notice the trap.
• Probability with a twist. The interviewer changes one condition after your first answer to see whether you understood it or memorized it.
• A research discussion. You walk through a past project and the interviewer pushes on everything that could have fooled you.
• Moderate coding. Load messy data, compute something, and explain what could make it misleading.

A typical regression pathology: you add a second predictor that is highly correlated with the first. Predictions might barely change. The two coefficients, though, can swing wildly and their standard errors balloon, because the model cannot tell which of the two variables deserves the credit. A candidate who stops at "R² went up a bit" misses what the question was for.

The research discussion is where researcher loops are won or lost, and you cannot cram for it. You need one project you know cold. Be ready to say what you hypothesized before looking at any results. Above all, know how many variants you tried before the one you are showing, because that is the question the interviewer is circling. The Sharpe-of-3 question in our researcher guide shows the kind of skepticism interviewers want to hear.

What Does the Same Setup Look Like in Each Loop?

Both loops can start from identical props. Here is one hypothetical setup, a single six-sided die, asked two ways.

Trader version: "Make me a market on how many sixes come up in 60 rolls." Fair value is 10. You quote around it, get traded against, and update. The interviewer cares about whether your first price came out in a few seconds and whether you moved when you should have.

Researcher version: "You roll a die 600 times and see 120 sixes. Is it loaded?" You expect 100. The standard deviation of the count is the square root of 600 × 1/6 × 5/6, which is about 9.1, so 120 sits roughly 2.2 standard deviations above expectation. That is suspicious but short of proof, and a strong candidate says so, then asks how many dice were checked before this one was flagged. If this die was the worst of twenty fair ones, chance alone would produce a result that extreme roughly one time in four.

The trader is graded on a fast, sensible price. The researcher is graded on whether they question where the number came from.

Which Track Should You Prep For?

Prep for the seat your applications point at. Splitting your time evenly between both "to be safe" is how candidates end up mediocre at both. Spend a base layer on shared probability, using the 20 quant interview questions that keep appearing as your core set, then commit.

If you are aiming for trader roles:

• Drill timed arithmetic daily, scoring accuracy as hard as speed.
• Practice market-making out loud with a friend or a mock tool, including being traded against.
• Learn the Kelly rule and why practitioners bet a fraction of it.
• Rehearse thinking aloud, since a silent correct answer often earns less credit than a narrated one.

If you are aiming for researcher roles:

• Work through regression and inference until you can explain each common pitfall in one sentence.
• Build one research project on a public dataset, with held-back data and a count of every variant tried.
• Practice a short coding exercise on messy data each week.
• Rehearse a two-minute project walkthrough that ends with what would have proved you wrong.

Both loops also include behavioral rounds that candidates underrate. The STAR method guide covers how to structure those answers.

How Can You Practice Each Path With PokeBot?

Start on the quant prep page and pick quant trader or quant researcher as your target role. Your resume is scored 0 to 100 against that role's hiring bar, and the skill-gap diagnosis measures readiness for that specific seat, so your prep targets one seat.

The Quant Interview Workbook has 56 questions with original step-by-step solutions. Trader candidates should go to the Brainteasers & Market-Making Games chapter. Researcher candidates have three chapters aimed at them: statistical inference and time series, financial machine learning, and portfolio and risk.

Then practice the spoken rounds in Interview Practice. Both signature rounds happen out loud, so reading solutions only gets you part of the way. Add your own prompts, such as a market-making setup or your research walkthrough, under "I have my own interview questions," and you get scored feedback on how specific and structured your answer is.

Your resume still decides whether you get the interview at all, so make sure it reads like the seat you want.

Score your resume free, create your PokeBot account