Jane Street Quantitative Researcher mock interview

Hybrid Trader+SWE track. Heavy probability + market-making games (Figgie). NO formal degree minimum — majority of researchers don't have PhDs. Interviewers WANT narrated reasoning and will tell you when you're wrong mid-problem.

  • ~$307K
    L1 first-year (Levels.fyi)
  • ~4 rounds
    Total loop
  • 4-8 weeks
    Process timeline
  • Free
    First mock

Why Jane Street QR sits between Trader and SWE — and why that matters for prep

Per Jane Street's official QR page (janestreet.com/quantitative-research), the official interviewing page (janestreet.com/join-jane-street/interviewing), Levels.fyi data, blog.janestreet.com 'What a Jane Street dev interview is like,' and aggregated candidate reports, JS QR is explicitly a hybrid track — not standalone. The official guidance: prep for both Trader (math, probability, market-making games) AND SWE (coding) tracks. ML Researchers (a QR sub-specialization) additionally work through 'realistic modeling problems' with the interviewer. There's no fixed round count published; reported structure is ~2-3 phone screens + Super Day, 4-8 weeks total.

The market-making game is uniquely weighted. JS uses Figgie (their own trading card game) and 'make me a market' problems on dice / decks. The graded dimension is how you reason under uncertainty in real time, update prices on observed trades, manage risk on your bid/ask spread — NOT whether you 'win.' Multi-part questions deliberately pivot assumptions mid-problem to see how you re-quote. Going silent when assumptions change is the standard fail mode.

Compensation is strong but discretionary. Per Levels.fyi: QR L1 $307K total, top reported $565K. TheWallStreetQuants reports new-grad QR total $350-600K (base $200-250K + discretionary year-end bonus). JS reportedly outbids Citadel for top new grads at QR level (per TeamRora). No PhD requirement (JS publicly states 'majority of researchers don't have PhDs') — strong MS candidates with math/stats/physics/CS/EE backgrounds break in. Putnam Fellow, IMO medal, or Codeforces Red rating override lower GPA / non-target school per efinancialcareers.

What the loop looks like

Round-by-round

  1. 01

    Recruiter screen

    Pre-onsite
    30 min

    Background, mathematical/coding interests, motivation. Per official JS page: Researcher is a mix of Trader and SWE prep — read both guides. No formal GPA/degree minimum mentioned.

  2. 02

    Technical phone screen × 2-3

    Pre-onsite
    60 min each

    Probability + modeling + coding hybrid. Math problems narrated out loud — interviewers grade on calibrated uncertainty and ability to handle pivoting assumptions mid-problem.

  3. 03

    Super Day — math/probability

    Onsite
    60 min

    EV, Bayes, conditional probability, MLE, hypothesis testing. Concrete problems like World Series Game 7 with two even teams, de Méré paradox, bee Markov chain stationary distribution.

  4. 04

    Super Day — market-making game

    Onsite
    60-90 min

    Figgie (JS's own card game) or 'make me a market' on a 6-sided die / 20-card deck EV. Tests real-time probability under uncertainty. Multi-part with assumption pivots.

  5. 05

    Super Day — coding / modeling

    Onsite
    60 min

    Language-agnostic — OCaml NOT required, no preference. Modeling-flavored problems (time series, experiment design, dataset generation) over LeetCode-style algorithms.

  6. 06

    Super Day — behavioral / fit

    Onsite
    60 min

    Humility, intellectual honesty, low-ego collaboration. How you handled critical feedback or ambiguity. Not 'leadership stories' — intellectual depth and curiosity gradations.

Question bank

Real interview questions reported by candidates

  • Probability

    "World Series Game 7 with two evenly-matched teams: probability the series goes to game 7? Walk through the binomial / negative binomial reasoning out loud."

    Source · tradermath.org Jane Street interview guide, Quora Jane Street problems
  • Brainteaser

    "25 horses, 5 lanes per race, no clock. Minimum number of races to determine the top 3 horses?"

    Source · Quora Jane Street interview problems
  • Probability

    "de Méré paradox: which is more likely — at least one 6 in 4 rolls of a single die, or at least one double-6 in 24 rolls of two dice?"

    Source · efinancialcareers.com Jane Street electronic trading interviews
  • Market-making game

    "Figgie game (JS's own card game): given the deck composition, make markets on the sums of cards. Quote prices, trade with the interviewer, update on new information."

    Source · medium.com/@mnshah0101 Jane Street interview question, blog.janestreet.com
  • Market-making game

    "Make me a market on the payout of a 6-sided die. Standard answer pattern: '3 at 4, 10 up' — bid 3, ask 4, with 10 contracts of size. Defend your spread."

    Source · efinancialcareers.com Jane Street electronic trading
  • Market-making game

    "20-card deck, 19 red and 1 black. Draw cards one at a time until you stop or hit black. EV of stopping at draw n? Now make a market on the EV."

    Source · efinancialcareers.com Jane Street electronic trading
  • Modeling

    "Modeling problem: design an experiment to test whether a new feature improves a key metric, given non-stationary user behavior and limited sample size."

    Source · interviewquery.com Jane Street, datainterview.com Jane Street QR
  • Behavioral

    "Tell me about a time you held an opinion you later realized was wrong. How did you update? What did the experience teach you?"

    Source · quantprep.io Jane Street interview
Common signals to fix

What gets you rejected at this level

  • Superficial probability intuition

    Per datainterview.com and quantprep.io: candidates who can't reason about EV/distributions/Bayes verbally — even if they could solve given pen and paper — fail. JS specifically grades on real-time reasoning under uncertainty. Practice probability OUT LOUD with someone listening, not silently.

  • Going silent when interviewer pivots assumptions

    Per efinancialcareers and interviewquery: JS interviewers deliberately pivot assumptions mid-problem to see how you re-quote/re-reason. Candidates who freeze rather than narrating their update flag missing the core skill JS hires for. Always vocalize: 'OK, with this new constraint, I'd update my estimate to…'

  • Can't articulate Bayesian updates verbally

    Bayes is foundational — not just for math rounds, but for market-making games where you update prices on new information. Candidates who know Bayes formula but can't apply it to 'I just saw a 6 — how does that change your market on the next die?' fail.

  • Generic 'I want finance' motivation

    JS culture values intellectual passion specifically. Generic finance interest reads as inauthentic. Reference specific JS work: blog.janestreet.com posts, OCaml open-source contributions, Figgie itself, or specific things you found interesting about the firm's research culture.

  • Ego / inability to admit a wrong answer

    Per quantprep.io and JS official interviewing page: 'asking great questions is more important than knowing all the answers.' JS interviewers explicitly tell you when part 1 is wrong before part 2 — they're testing whether you update gracefully. Candidates who defend wrong answers or get visibly rattled fail.

How Applr's AI mock interview tracks Jane Street's QR rubric

Applr's Jane Street QR mock simulates the hybrid format — probability brainteasers + market-making games + modeling problems graded together. The market-making rounds prompt 'make me a market' scenarios on dice / decks / Figgie-style game state, with explicit grading on real-time uncertainty reasoning and bid/ask spread defense.

Probability rounds prompt out-loud reasoning (matching JS's grading on verbal calibration, not silent solutions). Behavioral rounds prompt humility / wrong-answer-update scenarios — ego responses get flagged with rewrites that surface the JS-valued 'asking great questions over knowing all answers' framing.

Adjacent loops

Related practice

FAQ

Do I need a PhD for Jane Street QR?

No, NOT mandatory. Per official Jane Street page (janestreet.com/quantitative-research): no GPA or degree minimum, 70+ universities represented in the QR pool. Per efinancialcareers's coverage of JS employee perspectives: 'the majority of researchers at Jane Street don't have PhDs.' Strong MS candidates do break in. PhD is a plus but explicitly not required. What helps: math/stats/physics/CS/EE background, demonstrated probability/modeling chops. Putnam Fellow, IMO medal, or Codeforces Red rating widely cited as overriding lower GPA / non-target school. Publications matter for the ML Researcher track specifically; less for general QR.

What's the salary for Jane Street QR?

Per Levels.fyi (May 2026): QR L1 $307K total ($256K base + $51.7K bonus, $0 stock), L2 $210K, median $300K, top reported $565K. Per TheWallStreetQuants industry guide: new-grad QR base $200-250K, total $350-600K. Per TeamRora: JS currently outbids Citadel for top new grads at the QR level. Bonus is FULLY DISCRETIONARY year-end, driven by desk PnL + individual model impact — no public signing bonus figures. Compare: Citadel QR new-grad ~$400-651K, Two Sigma QR ~$230K (closer to Big Tech), JS QR $350-600K.

What is the Figgie game actually like?

Per medium.com/@mnshah0101 and blog.janestreet.com: Figgie is JS's own trading card game. Decks have 4 suits but with non-uniform distributions (e.g., 12 spades, 10 each of others); one suit (the 'goal suit') has more cards than expected. Players make markets on the sums of cards in their hands and the goal suit. The interviewer plays with you, quoting prices and trading. The graded dimension is how you reason under uncertainty in real time, update on trades observed, manage risk on your bid/ask spread — NOT whether you 'win.' Practice mathematical games and probability under time pressure before the loop.

What language should I code in?

Per official JS page: language-agnostic for QR. OCaml is NOT required — most hires don't know OCaml coming in. JS explicitly discourages learning OCaml just for interviews ('please don't use OCaml just because you think it will make us happy'). For QR specifically, Python is most common (modeling work, time series, dataset manipulation). Use whatever language you're most fluent in. The graded dimension is clean modular thinking, not language choice.

How does QR differ from Trader at Jane Street?

Trader: heavy mental math, probability, market-making games, NO coding required. SWE: coding-only, any language, no math. Researcher (QR): probability + modeling + coding hybrid — JS describes the work as 'overlapping both trading and software engineering.' QR work is more strategy/model development; Trader work is more execution and live decision-making. ML Researchers (a QR sub-track) work through realistic modeling problems with the interviewer. If you have strong math + want to code, QR. Strong math + want to trade live, Trader. Strong code + want minimal math, SWE.

Does Jane Street sponsor visas for international students?

Yes for QR roles in NYC and London offices. Visa coordination varies by office and timing — confirm with recruiter early. Hong Kong office also sponsors local visas. JS is a strong sponsor — international students with US/UK work eligibility have a clear path.

Ready to practice for Jane Street?

Start free →