All articles Guest Post

What I Went Through in the Jane Street Quant Trader Interview

Guest post by Jeff Personal account ~10 min read
Draft A personal account, based on a talk, lightly transcribed and paraphrased with AI. It’s one person’s experience from a couple of years ago, not an official or definitive guide — the process changes, so take it as a starting point, not gospel.

Hey everyone, I’m Jeff. Quick disclaimer before anything else: I don’t work at Jane Street. A couple of years ago I was lucky enough to get a Jane Street internship offer, but I ended up taking a different full-time offer instead. The reason I passed it up: the return rate is rumored to be low — I’d heard somewhere around 30% — and since almost all of Jane Street’s full-time hiring runs through the internship, taking the intern spot basically meant re-rolling the dice on a 1-in-3 conversion. I already had a full-time offer in hand that I actually wanted, so that wasn’t a re-roll I felt like making. I’m in the industry now, working at another firm in the space.

Thanks to Stanley for inviting me to share this. I don’t have the definitive playbook — I’ll just walk through what I went through and what stuck with me. For context: I interviewed for quant trader, did about two technical rounds before my Super Day, and got the offer. So this covers the front half of the funnel and the Super Day itself, which is the part everyone wants to hear about anyway.

If you want the structured version of all this — the rounds laid out with the actual reported questions and practice for each — QuantVault keeps a live Jane Street funnel. I’ll point at the relevant pieces as I go.

First, what Jane Street is actually like

A few things I had wrong going in, that turned out to matter:

The collaborative thing is real — and I felt it in the interview. The Jane Street interview was the best one I’ve ever sat in. The people across the table were sharp, and they were working the problem with me instead of just clocking my mistakes. Most other places I interviewed at didn’t feel anything like that. It also shapes how you should behave in the room (more on that below).

The comp is flatter than people expect. This part is second-hand — I have a couple of friends at the firm, and what they tell me is the bonus structure is pretty flat. Jane Street pays extremely well, but they trade as a firm, not as a pile of individual books, so you won’t dramatically outpace your peers off one hot year. If your whole goal is to maximize personal income variance off your own outsized alpha, other places have higher tails. For most people, though, it’s very competitive on total comp.

Almost everything flows through the internship, which is the part that scared me off. Friends who did it told me the same two things: it’s brutal, and not many interns convert. The offer just gets you in the door; staying in is a whole second fight. I didn’t do it myself so I can’t speak to it first-hand, but that low conversion is the whole reason I went full-time somewhere else.

The technical rounds (I had two)

Before the Super Day I did two technical phone/video rounds. Other people I know did anywhere from one to four — it varies. Here’s what stood out.

The questions felt hand-written. A lot of them are written by the traders sitting across from you, so grinding a list of “known” problems didn’t do much for me. What helped was intuition: being fast enough on the fundamentals to handle a twist I’d never seen.

What I got hit with, roughly:

  • Probability, expectation, and combinatorics — usually with a clever twist that changed the “obvious” approach
  • Game-theory reasoning — not “state the theorem,” but “reason about the equilibrium out loud”
  • Brain teasers that looked like one kind of problem and were actually another
  • A bit of behavioral woven in, with real follow-ups

The single biggest thing I’d tell you: talk while you think. I almost blew a round going quiet. These are conversations. Going quiet for two minutes and then announcing the right answer is worse than narrating a slightly messy path to it. They want to see the reasoning live.

What I’d actually do is decompose it out loud, like this:

  • Say the goal first: “I want the expected number of rounds before…”
  • Then figure out what it depends on: “so I need the probability of X at each step…”
  • Then grind through the pieces — condition on cases, whatever the problem needs

Don’t narrate every dead end — just show the spine of the logic as you build it. I had to drill this on purpose, because my default was to go quiet and think. If you change one thing, change this.

On game theory specifically — I made sure I could explain each of these from scratch, in plain English, without drawing a payoff matrix: pure and mixed Nash equilibria (and when mixed ones exist), dominant strategies and iterated elimination, the standard game forms (prisoner’s dilemma, coordination, stag hunt), and the auction formats (first-price, second-price, all-pay). If you need the matrix to reason about a simple symmetric 2-player game, you’re not fluent enough yet. That was my bar.

How I prepped this part: I drilled probability and EV until the approach was automatic (I used the probability sets for volume), and I practiced everything out loud — I’d record myself and watch it back, which is miserable, and that’s what actually fixed me. To calibrate on the firm’s style, I’d filter to Jane Street-tagged problems and rounds.

The Super Day

Clearing the phone rounds got me a Super Day — a full day onsite. They seemed pretty generous about inviting people; the vibe was “we’d rather meet you in person than cut you early over the phone.” They paired me with interviewers whose background looked like mine, which made the conversations feel fair.

It splits into a morning and an afternoon, and they’re very different days.

Morning: the market-making games

The morning was entirely market-making games — same format whether you’re going for trader or researcher. You get chips (your capital), there’s some random process (dice, cards, a constructed scenario), and you quote two-sided markets on some quantity over many rounds. Sometimes several things trade at once.

This was the part I was most nervous about and, in hindsight, the part where prep mattered most, because you can’t bluff your way through a live quote. Here’s the routine I ran on every market:

1. Estimate fast, and say how. The number didn’t need to be perfect, but I had to get there quickly and explain it: “treating this as a sum of three independent uniforms, EV is about X because each has mean Y…” I was talking them through how I got there, not just dropping a number.

2. Think about variance, not just EV. This is where I saw people lose it — they’d anchor on the mean and forget that it’s the width of the distribution that sets your spread. High variance → quote wider to protect against adverse selection. Tight → you can be aggressive. The interviewers were clearly watching for whether I accounted for variance at all.

3. Check your downside before you quote. Before every market: what’s my max loss if I get hit on both sides at my prices, and how big is that vs my chip stack? My rough rule was to keep single-trade downside to maybe 70–80% of the stack at most. You’re just trying to last long enough for the edge to show up. Blow up the stack on one badly-sized trade and you’re done. I said this out loud every time.

4. Hunt for hedges and arbitrage. When multiple things trade together, look at how they relate. If I was long something correlated with something I was short, my real risk was smaller than it looked — so I could size up. And if a clean arb existed, I’d take the whole thing on the spot.

5. Say all of it. The market-making game is at least as much a talking test as a math test. I narrated where I was quoting, what edge I thought I had, what worried me, how I was sizing. A good trade I made quietly counted for less than a so-so one I talked through.

The only way I actually got comfortable was repetition — making markets on simple random processes over and over. QuantVault has scored mocks of these exact games, which is the closest thing to the real thing I’ve found:

Three-Dice Min → Make a market on the minimum of three hidden dice — the canonical Super Day game. Make a Market → Quote a bid/ask under a min-spread rule and get filled by a live counterparty. Beat the Odds → A board of odds under a timer — find the +EV bets and the arbitrage before it closes. Card Sum Trader → Price the running sum of a dealt hand and manage inventory as cards turn over. Hedge the Sum → Same sum, now with a hedging instrument — lay off risk instead of guessing. RGB Counts Product → Market the product of red/green/blue counts — nastier distribution, fatter tails.

If you only play one before your loop, play Three-Dice Min until the routine above (estimate → variance → downside → hedge → quote, all out loud) feels automatic.

Lunch

If the morning went well, you get lunch with traders. It’s semi-formal — partly social, partly a soft read on culture. One thing I noticed: a few people who struggled in the morning quietly left after lunch. So if you make it to the afternoon, you’ve already cleared the hardest part.

Afternoon: coding, data, and behavioral

The afternoon was more varied and more tailored to my background — more games in different formats, a coding component, some data work, and behavioral questions that actually dug in.

The coding round surprised me more than anything. There was no LeetCode-style “implement this in 20 minutes against a hidden test suite.” I got a terminal and an open-ended problem and was asked to write a function to solve it. The twist: you write your own test cases. No hidden checker, no expected output — you decide what edge cases matter and how to know you’re right.

The interviewer basically acted like a teammate — hinting, pushing back, working through it with me. The right mode is “let’s solve this together,” and if you’re stuck, say so and ask. What got me closest in practice was the Interview mode on QuantVault’s coding problems — a blank editor and a real terminal where you write and run your own tests, no checker. Practice solving open-ended and defending your own test design out loud.

For data: pandas is the standard now (they used to be more Excel-heavy). Be comfortable with filtering, groupby, merges, and basic stats.

Behavioral was tailored. They’d clearly read my CV and dug into specific projects, ambiguous decisions, and how I work on a team. I’d prep 3–5 concrete stories — context, what you did, outcome, and what you’d change — not generic “tell me about a time” answers.

One thing that ties it all together

The mindset that I think actually got me the offer: treat the interviewer as a teammate from the very first question. Don’t try to perform. Solve it with them. If you’re stuck, say so and ask for their read instead of going silent; if they nudge you toward something, actually take the nudge; if you’ve only got half an idea, put it on the table and build from there. From what I gathered, this isn’t just fit-signaling — it’s literally how the place works day to day, and they use group games in the internship to keep testing it. The more naturally you operate this way, the more convincing it is.

If I were prepping again, here’s my short list

  • ☐ Probability + EV + combinatorics till it’s reflex — speed and intuition over memorized lists
  • ☐ Game theory I can explain from scratch, no matrix needed
  • ☐ The market-making routine: estimate → variance → downside → hedge → quote
  • ☐ Variance intuition specifically — always ask how wide the distribution is, not just where it’s centered
  • ☐ Open-ended coding where I design the tests; comfortable in pandas for the data stuff
  • ☐ Speak-while-thinking — I drilled this daily, recording myself
  • ☐ 3–5 behavioral stories I’d actually rehearsed

That’s what I remember — one guy, a couple years out of date, on the quant trader track. The hard part about Jane Street is also the fair part — everything they test, you can build by just repping it. Good luck, and thanks again to Stanley for the nudge to write this down.

— Jeff

Want to drill the rounds I described?
The technical rounds, the Super Day market-making games, and the open-ended coding round are all mapped out with real questions and playable mocks.
Open the Jane Street funnel