Jane Street Interview Process 2026: Mental Math, Trading Games & OCaml

Last updated: August 2026

TL;DR

Jane Street’s interview process runs an application and a timed online assessment, then one or two phone or video rounds of probability, estimation and (for engineers) live coding, then a final in-person day of roughly three to six long sessions covering mental math, trading games, coding, and a conversation about your past work. Most candidates describe 3-6 weeks from first contact to decision, and campus candidates move faster because the firm reviews applications on a rolling basis. Drill probability and estimation out loud in OphyAI Interview Practice, rehearse the evolving-coding format with OphyAI Coding Interview, and use OphyAI Interview Copilot for mock rounds on Zoom, Teams, or Meet. The final day is in person, so treat any tool as practice-only.

Quick Answer: Jane Street Interview Process

Stage	What to prepare
Application and resume review	A resume that shows something you went deep on: a research project, a competition, a hard piece of software. No finance background needed.
Online assessment (mostly trading and research tracks)	Timed probability, mental arithmetic, and logical reasoning under time pressure.
First phone or video round	Rapid-fire expected-value questions, estimation, and (for engineers) a live coding problem on a shared editor.
Second technical round	A deeper problem: a trading or market-making game for traders, an extended coding problem for engineers.
Final in-person day	Three to six sessions back to back, usually with two interviewers each, on math, games, coding, and your prior work.
Offer and decision	Base, sign-on and a discretionary first-year bonus; ask how the bonus is determined before you respond.

Action Plan: Prepare for Jane Street by Round

Round	What Jane Street tests	What to do before the interview
Online assessment	Numeracy under a clock, not exam algebra	Drill timed arithmetic (percentages, fractions, powers of two) and 40-60 expected-value questions until the setup is automatic
First phone round	Whether you can price uncertainty out loud and take a hint	Practise narrating probability answers, including the moment you notice you are wrong
Trading game	Bid-ask reasoning, updating on new information, position sizing	Play market-making games with a friend: someone quotes, someone hits, then both explain the edge
Coding rounds	Clean, extendable code in your best language, plus questions on an underspecified prompt	Take one medium problem and add three new requirements to it without rewriting from scratch
Past-work conversation	Depth on something you actually built or researched	Pick one project and be able to defend every design decision three layers down
Final day, all rounds	Stamina and how you behave when a problem gets harder	Run one full mock day of back-to-back sessions so fatigue is not new information on the day

If you only have one week, spend it on two things: timed mental arithmetic, and saying probability answers out loud with your reasoning exposed. Jane Street interviewers score the reasoning, and a silent candidate who arrives at the right number scores worse than a candidate who thinks clearly and gets corrected.

What Makes Jane Street Different

Jane Street is a proprietary quantitative trading firm founded in New York around 1999-2000 (published sources differ on whether to date it from incorporation or first trading). It trades ETFs, options, equities, bonds and cryptocurrencies across more than 200 venues worldwide, with reported monthly equity volumes in the trillions of dollars. Roughly 3,000 people work in the New York headquarters and in London, Hong Kong, Singapore and Amsterdam, which makes it small relative to the banks it trades against and unusually dense in technical staff.

Several characteristics shape what interviewers look for:

No finance background required, and none expected. Jane Street’s own campus materials say there are no real requirements to apply and that they will teach you the finance. That is not a courtesy line: trading interviews use probability and games, not discounted cash flow models, and candidates from mathematics, physics, computer science and competitive programming backgrounds are common.
OCaml everywhere, but not in the interview. Jane Street is the largest commercial user of OCaml and runs research tools, trading systems and infrastructure on it. The firm states you do not need to know OCaml, that most engineers arrive without functional programming experience, and that there are no bonus points for using a functional language in the interview. Use the language you know best.
Questions over answers. Jane Street says asking great questions matters more than knowing all the answers, and its coding rounds are deliberately underspecified so the first thing you do is ask. Candidates who start typing immediately lose a scored dimension.
Collaborative, two-interviewer rounds. Most final-day sessions have two interviewers. Candidates consistently describe it as a joint working session rather than an interrogation, with hints offered and expected to be used.
Calibrated uncertainty is the skill. Traders are hired to be right about distributions, not about single numbers. Interviewers reward candidates who say what they do not know, price it anyway, and update cleanly when new information arrives.

Many candidates rehearse Jane Street-style rounds with OphyAI Interview Copilot on Zoom, Teams, or Meet to keep their reasoning audible and structured. Jane Street’s final round is in person and the firm has not published a policy on outside tools, so treat Copilot as a practice tool and sit the real rounds unassisted.

Interview Process Overview

The loop varies by track, and the trading and engineering paths diverge after the first screen. The shape below is what most candidates describe.

Stage	Format	Duration	Timeline
Application and resume review	Online	-	Week 1
Online assessment (trading and research tracks; some engineering candidates)	Timed online test	60-90 minutes	Week 1-2
First phone or video round	Video, often with a shared editor for engineers	45-60 minutes	Week 2-3
Second technical round	Video	45-60 minutes	Week 3-4
Final day	In person, 3-6 sessions, usually two interviewers each	About 60-70 minutes per session	Week 4-6
Decision and offer	Phone and written	-	Week 5-7

The first contact is fast. Jane Street says a recruiter usually gets in touch within a few days of your application and will tell you what the next step is. Campus hiring runs on a rolling basis, so applying early matters more than polishing your resume for another month.

The online assessment is the main filter for trading and research candidates: timed probability, arithmetic and logical reasoning, typically over 60-90 minutes. Rehearse it by talking through timed problem sets in OphyAI Interview Practice and reading the transcript for the moments where you stalled.

Phone and video rounds are short, dense, and conversational. Traders get rapid-fire expected-value and estimation questions. Engineers get a live coding problem on a shared editor, usually on Zoom with a collaborative editor such as CoderPad. Both tracks are scored on the reasoning you say out loud.

The final day is in person. Engineers describe three to four long coding sessions of roughly 70 minutes each; traders describe four to six sessions mixing mental math, probability, trading games, and past work. Two interviewers per session is normal. Run one full mock day with a friend using OphyAI Interview Copilot so that the fifth hour is not the first time you have thought hard while tired.

The Mental Math Test and the Trading Games

Two assessments define the trading track, and both are unusual enough that generic interview prep does not transfer.

The mental math test. Arithmetic under a hard clock: percentages, fraction-to-decimal conversion, awkward multiplications, quick estimation. Speed matters, but the underlying test is numeracy and comfort with approximation rather than exact calculation. A useful daily drill: take any three-digit multiplication, produce a one-second estimate, compute the exact answer, and note the size of your error.

The trading games. These are interactive market-making exercises. A typical version: the interviewer describes a random variable (the sum of dice, the number of a certain letter on a page, the outcome of a game with hidden information) and asks you to quote a two-sided market. They then hit your bid or lift your offer, reveal a piece of information, and ask you to requote. What is being scored:

Dimension	What good looks like
Initial quote	A width that reflects your actual uncertainty, not a token spread
Updating	Moving your market in the right direction and by a defensible amount when information arrives
Adverse selection	Noticing that the counterparty trading with you is itself information
Position awareness	Knowing what you are long or short and what that does to your next quote
Composure	Being wrong out loud and recovering, rather than freezing

Critical advice: the failure mode is not bad arithmetic. It is quoting a narrow market to look confident, getting run over, and then defending the original quote. Interviewers are explicitly interested in how you handle adversity and incorporate feedback, so treat a bad fill as information and say what it changed.

Role-Specific Breakdowns

Quantitative Trader

The trading loop is probability-first. Expect the online assessment, a first round of rapid-fire expected-value and estimation questions, a second round built around a trading or market-making game, and a final day of four to six sessions.

What is tested. Expected value, conditional probability, Bayesian updating, simple game theory, estimation (the Fermi-style “how many X in Y” family), and your behaviour when a game goes against you. Interviewers add complications mid-problem to see whether your reasoning survives them.

How to prepare. Work probability problems out loud, not on paper. Practise saying the setup (“I am pricing the expectation of X given Y, so my fair is around Z, and I would quote around it with a width of W”) before you compute. Drill it in OphyAI Interview Practice in case format and use the per-answer feedback to cut the hesitation out of your explanations.

Software Engineer

Jane Street’s engineering interview is a coding interview, not a system design interview. Candidate accounts consistently report no separate system design round in the standard loop.

Phone or video screen (45-60 minutes). One live problem on a shared editor. Expect the problem to be underspecified on purpose.

Final-day coding rounds (about 70 minutes each, three to four of them). Each round is usually one problem that grows: you solve it, then the interviewer adds a requirement, then another. The skill being tested is whether your first design can absorb changes without a rewrite.

What interviewers look for. Code that looks like production code rather than pseudocode with placeholders. Correctness you can argue for. Comfort with the standard data structures and APIs of your chosen language. Narrated debugging: when something breaks, say what you are checking and why instead of scanning silently.

Senior candidates. Expect an additional conversation about a system you built, where the questions go into your actual decisions rather than a generic design prompt.

Use OphyAI Coding Interview to rehearse the evolving-requirements format: take a problem, get a streamed approach, then re-implement it yourself and add two new constraints without starting over. For general fundamentals, our coding interview examples with solutions covers the patterns these rounds assume you already have.

Quantitative Researcher and Trading Technology

Research roles blend the two tracks: probability and statistics depth, plus coding in Python or the language you are strongest in. Expect questions about how you would test whether a signal is real, what you would do about overfitting, and how you would size a bet given an edge and a variance estimate. Trading technology roles look more like the software engineering loop with a heavier emphasis on latency and correctness under load.

Rehearse explaining one research result end to end in OphyAI Interview Practice in technical format, including the part where you say what would have falsified it.

Intern and New Graduate

Jane Street’s campus pipeline is large and runs early: applications for the following summer typically open in the preceding August and are reviewed on a rolling basis, with the most popular offices filling first. The internship interview is close to the full-time loop with less weight on prior work history, because interns are not expected to have a track record.

Stages. Application, online assessment, one or two video rounds, and a final day. Candidate reports on Glassdoor describe a comparatively fast turnaround for intern roles, on the order of a couple of weeks from first contact.

What to bring. One thing you went deep on. A competitive programming record, a research project, a game engine, a trading bot you actually ran: the specific artifact matters less than your ability to answer three follow-up questions about it.

Use OphyAI Interview Practice to rehearse the project walk-through until you can explain it to a non-specialist in ninety seconds and to a specialist for ten minutes.

Common Questions with Frameworks

1. “You roll two dice. Make me a market on the sum.” (Trading Game)

Approach: State your fair value first (7), then choose a width that reflects your uncertainty about the rules rather than your confidence. Quote, then react. When the interviewer trades with you, say out loud what that tells you: if they lifted your offer instantly, your offer was probably too low. Requote and explain the move. The score is in the narration, not the number.

2. “How many piano tuners are there in Chicago?” (Estimation)

Approach: Decompose into a chain you can defend: population, households, share with pianos, tunings per year, tunings per tuner per year. State each assumption as you make it and flag which one your answer is most sensitive to. Finish with a range and the direction you think you are biased. Precision is not the point; auditable structure is.

3. “Write a function that groups events into sessions.” (Coding, underspecified on purpose)

Approach: Ask before you type. What is an event, is the input sorted, how large can it get, what is the session gap, what should happen at boundaries. Then write real code in your best language. When the interviewer adds “now sessions can overlap across users”, your first design should bend rather than break. Say what you would test.

4. “I flip a coin until I see two heads in a row. What is the expected number of flips?” (Probability)

Approach: Set up states rather than reaching for a formula: no heads yet, one head, done. Write the expectation for each state in terms of the others and solve. Narrate the setup before the algebra so an interviewer can correct a wrong model early instead of watching you compute the wrong thing correctly.

5. “Tell me about something you built or researched that you are proud of.” (Past Work)

Approach: Pick something real and go deep rather than broad. Use the STAR method loosely, but expect the questions to leave the story and enter the design: why this data structure, what broke, what you would do differently. Vagueness at layer three is the common failure.

6. “Why Jane Street?” (Motivation)

Approach: Answer honestly and specifically. Good answers name the work: markets as a fast feedback loop, the OCaml codebase, the fact that the firm is small enough that your decisions have visible consequences. Weak answers describe compensation or repeat the phrase “intellectually stimulating” without a referent.

For broader practice on the non-technical half of the loop, see our guide to common interview questions and answers.

Culture Fit: Collaboration and Calibrated Uncertainty

Collaboration is the format, not a value statement. Two interviewers per session, hints offered freely, and problems designed to be worked jointly. Candidates who treat the interviewer as an adversary to be impressed usually score below candidates who treat them as a colleague to think with.

Saying “I do not know” is a positive signal, once. Traders are paid to distinguish what they know from what they are guessing. Say which is which, then commit to a number anyway. What does not work is refusing to guess.

Questions are scored. Jane Street states plainly that asking great questions matters more than having all the answers. In coding rounds this is literal: the prompt is incomplete and the clarifying questions are part of the evaluation.

What interviewers screen for: numeracy under pressure, reasoning you can hear, willingness to be corrected mid-problem, depth on one thing rather than surface familiarity with many, and the absence of finance jargon deployed to sound like an insider.

Compensation Overview (2026 Estimates, USD)

All figures below are candidate-reported, sourced from Levels.fyi and third-party compensation trackers and last checked in August 2026. Jane Street does not publish salary bands. Its bonus is discretionary and performance-linked even in the first year, which is why reported ranges are unusually wide.

Role	Base Salary	Total Compensation (Base + Sign-on + Bonus)
Software Engineer (L1, new grad)	Candidate-reported base is a minority of the package	~$382,000 median reported
Software Engineer (all US levels)	Varies by level	~$420,000 median reported
Backend Software Engineer (senior levels)	Varies by level	Reported range roughly $338,000 to $650,000+
Quantitative Trader (new grad)	Reported first-year packages	Commonly reported at $400,000 to $600,000+
Trading or software intern	Monthly stipend	Third-party guides report roughly $25,000-$30,000 per month in New York and about GBP 15,000-18,000 in London, plus housing

Treat all of these as candidate-reported medians and ranges rather than published bands, and note that the bonus dominates the package: the low end of a reported range is a quiet first year and the high end is a strong one. Ask your recruiter how the discretionary bonus is determined, what the sign-on structure is, and whether any part of the first-year number is guaranteed. Our salary negotiation guide covers how to have that conversation without anchoring yourself low.

Preparation Timeline: 4-6 Weeks

Week	Focus	Activities
1	Baseline and arithmetic	Test yourself on timed mental math daily. Establish your error rate before you try to get faster. Read the firm’s own campus and engineering interview pages so your expectations match theirs.
2	Probability and expected value	Work 40-60 problems out loud, covering conditional probability, Bayesian updating, expectation via states, and simple games. Say the model before the algebra.
2-3	Estimation and games	Do one Fermi estimate a day with a stated sensitivity. Play market-making games with a friend, alternating who quotes. Track when you were run over and why.
3-4	Coding (engineers) or research depth	Practise one evolving problem per session: solve, then add two requirements. Write real code in your strongest language, no pseudocode. Use OphyAI Coding Interview for approach checks, then re-solve unassisted.
4-5	Past work and mock rounds	Prepare one project you can defend three layers down. Run mocks in OphyAI Interview Practice and cut filler from your explanations.
5-6	Full-day rehearsal and logistics	Run one back-to-back mock day to build stamina. Prepare questions for each interviewer. Confirm travel and ID for the in-person final. Sleep before the day, because the fourth session is where tired candidates lose points.

Common Mistakes

Training for speed before accuracy. Chasing arithmetic tricks without fixing your error rate makes you faster at being wrong. Establish accuracy first, then compress.

Quoting a tight market to look confident. A narrow spread that you cannot defend invites the interviewer to trade against you and then watch you rationalise. Width is information about your uncertainty, and honest width scores better than false precision.

Typing before asking. The coding prompt is underspecified deliberately. Starting with implementation skips a graded step.

Pseudocode and placeholders. Jane Street expects code in a real language that resembles what you would write at work, including the parts that are tedious.

Learning OCaml the week before. The firm awards no bonus points for functional languages and recommends the language you are most fluent in. An interview is a bad place to be a beginner.

Going quiet when a problem gets hard. Silence removes the thing being scored. Narrate the dead ends.

Prepare for Jane Street with OphyAI

Jane Street’s loop rewards a specific combination: fast, honest numeracy; reasoning that is audible; and code that survives a change of requirements. All three improve with rehearsal, and none of them improve by reading more solutions.

Use OphyAI Interview Practice for voice or text mock interviews in technical, case and behavioral formats, with a transcript and per-answer feedback so you can see exactly where your explanation lost the thread. Use OphyAI Coding Interview to rehearse an evolving coding prompt, then re-solve it with nothing open. Use OphyAI Interview Copilot during mock rounds on Zoom, Teams, or Meet to keep your structure tight, and sit the real interviews unassisted. Start practicing →

Start Your Jane Street Application

Ready to apply? OphyAI can help at every stage:

Search for open roles at Jane Street and similar firms with AI-powered job matching
Generate a tailored cover letter that highlights your fit for the role, plus follow-up emails and thank-you notes for after your interviews
Track your application status alongside every other role you’re pursuing

Pair these with Interview Copilot for structured mock interviews, or practise first with OphyAI Interview Practice.

For product details, see Interview Copilot.

Frequently Asked Questions

How many rounds are in the Jane Street interview process?

Most candidates describe four to seven contacts in total: an application review, a timed online assessment for trading and research tracks, one or two phone or video technical rounds, and a final in-person day made up of three to six back-to-back sessions. Engineering candidates commonly report three to four coding sessions on the final day, each around seventy minutes and usually with two interviewers. The exact count varies by office and track, so ask your recruiter to lay out your specific loop when they first call.

Do I need a finance background to interview at Jane Street?

No. Jane Street’s own campus materials state that there are no real requirements to apply and that the firm will teach you the finance you need. Trading interviews test probability, expected value, estimation and game play rather than accounting or valuation, and successful candidates commonly come from mathematics, physics, computer science and competitive programming backgrounds. What matters is demonstrable quantitative reasoning and evidence that you have gone deep on something difficult.

Do I need to know OCaml to get a software engineering job at Jane Street?

No. Jane Street is the largest commercial user of OCaml and runs most of its systems on it, but the firm states that you do not need to know it, that most of its engineers arrived without functional programming experience, and that interviews award no bonus points for using a functional language. Use the language you are most fluent in, because interviewers care about clean, correct, extendable code and clear communication rather than which language produced it.

How hard is the Jane Street mental math test?

Candidates consistently describe it as demanding, but the difficulty comes from the clock rather than the mathematics. The content is arithmetic, percentages, fractions and estimation, not advanced algebra. What is really being assessed is numeracy and an intuitive feel for magnitudes, so a candidate who estimates well and refines quickly usually outperforms one who tries to compute everything exactly. Daily timed drills over three to four weeks move most candidates measurably.

How long does Jane Street take to respond after an interview?

Jane Street says a recruiter typically contacts you within a few days of applying and will tell you what to expect next, and candidate reports describe a comparatively quick process overall, often three to six weeks from first contact to decision. Intern candidates on Glassdoor report some of the fastest turnarounds, on the order of a couple of weeks. Campus hiring is rolling, so applying early in the cycle affects both your odds and your timeline more than at employers with fixed windows.

Can I use an AI interview assistant during a Jane Street interview?

Treat outside tools as not permitted unless your recruiter explicitly says otherwise, and ask if you are unsure. Jane Street’s final round is conducted in person, which makes any on-screen assistant impractical, and the firm has not published a policy covering its earlier video rounds. The productive use of Interview Copilot and Interview Practice here is mock rounds before the real loop, where feedback on your structure and pacing is exactly what the final day will test.

What should I do if I get a trading game question wrong?

Say so and update. Trading games are designed to move against you: the interviewer will trade on your quote and then reveal information, and the scored behaviour is whether you notice, adjust in the right direction, and explain the adjustment. Defending an indefensible quote is the failure mode; changing your market with a stated reason is the pass. Interviewers describe themselves as interested in how candidates handle adversity and incorporate feedback, so a visible recovery is worth more than an unbroken run.

Sources and verification notes

All sources were checked in August 2026 through published guides and candidate-report aggregators. Jane Street publishes guidance about applying and interviewing but does not publish round counts, assessment lengths, scoring rubrics or salary bands, so every number in this guide that describes those is the range candidates most often report, labelled as such. Direct fetches of first-party pages were not possible from the environment this guide was written in, so first-party statements below are cited to the pages that carry them and were cross-checked against at least two independent write-ups before use.

Jane Street, Interviewing: the firm’s own description of what to expect, the emphasis on collaborative problem solving and asking questions, and recruiter response timing.
Jane Street, Preparing for a Software Engineering Interview: use your strongest language, no bonus points for OCaml, write real code rather than pseudocode.
Jane Street, Technology and OCaml.org success story: the OCaml stack and its scope across research, trading and infrastructure.
Wikipedia, Jane Street Capital: founding, offices, headcount of roughly 3,000, asset classes and venue count, and reported trading revenue figures.
Levels.fyi, Jane Street Software Engineer and Jane Street L1: candidate-reported United States medians and ranges, checked August 2026.
Exponent, Jane Street Software Engineer Interview Guide and Jane Street SWE Intern Guide: screen format, final-day session count and length, the evolving-problem structure, and the absence of a separate system design round.
Glassdoor, Jane Street Quantitative Trader interview reports and Jane Street Intern interview reports: candidate-reported round formats, question types and turnaround times. Self-selected reports, so read them for pattern rather than precision.
techinterview, Jane Street Interview Guide 2026: mental math test, probability brainteasers and trading games as the trading screen rather than algorithm rounds.
Quantt, Jane Street Interview Guide and Jane Street Internship Guide: stage-by-stage trading loop, intern application timing and reported stipends. Third-party, treated as typical rather than official.