A first-year quant trader at Jane Street can clear $400,000 to $700,000 in total comp, and the base salary is the least interesting number in that range. The interesting number is the bonus: discretionary, tied to how the firm and your desk did, and large enough that everything else rounds to noise.
That gap between base and total is what people miss when they search ‘Jane Street salary’ and get back a single tidy figure. Glassdoor and Levels.fyi will hand you an average, and an average hides the only thing worth knowing, which is how the pay is built and why a proprietary trading firm can pay a 22-year-old more than a staff engineer at most public tech companies makes in a strong year.
The base is a floor, and the firm keeps raising it
For new grad traders, the base has been climbing fast. The 2025 class saw bases in the $225,000 to $300,000 band; reporting on the 2026 cohort puts the floor nearer $200,000 to $250,000, with stronger offers above that. Jane Street resets entry-level base upward almost every cycle, mostly to win bidding wars with Citadel Securities, Hudson River Trading, and Optiver over the same tiny pool of graduates.
Interns get the same math. Jane Street pays most of its interns the annualized equivalent of roughly $300,000, and it applies that rate across quant traders, quant researchers, machine learning engineers, FPGA engineers, and network engineers. A ten-week summer at that rate lands somewhere near $60,000 before any housing stipend. That is not a misprint, and it is why the firm tops the ‘highest-paid intern’ lists that circulate every spring.
The base is built to be a floor, not the destination. Once you are full-time it barely moves year over year. Everything that produces the headline numbers happens in the bonus.
The bonus is the whole game
Jane Street trades its own capital. No outside investors, no client assets, no management fee skimmed off someone else’s money. The dollars that land in your account come straight out of the trading profit the firm booked that year. That one fact explains the entire structure. In a wild, high-volatility year the firm prints, and bonuses balloon. In a quiet year they compress, and the people who joined expecting last year’s number get a quick lesson in how prop pay actually behaves.
For a strong first-year trader, the bonus alone can match or beat the base, which is how a $250,000 base becomes a $500,000-plus total. Three or four years in, running real risk on a desk, the bonus is most of the pay and the gap between a median performer and a top one gets very wide. Two traders who started the same Monday can be a factor of two or three apart by year four. Nobody publishes that distribution, because the firm doesn’t, and the people at the top of it have no reason to.
What the 2026 numbers actually look like
Treat everything below as a range with a fat error bar. Bonus is discretionary, self-reported data skews high, and one great or terrible firm-wide year shifts the whole picture.
| Role and stage | Approx. base | Approx. total comp (2026) | What drives the variance |
|---|---|---|---|
| New grad quant trader | $200K–$300K | $400K–$700K (year one) | Signing bonus and any first-year guarantee inflate year one; desk PnL after that |
| New grad quant researcher | $200K–$250K | $350K–$600K | Tied to firm performance more than to a single desk |
| New grad SWE / quant dev | $175K–$225K | mid-$300Ks (Levels.fyi self-reported) | Bonus is real but less tied to PnL than trading |
| Intern, any technical track | ~$300K annualized | ~$60K for a 10-week summer | Flat rate across roles |
| Trader, 3–4 years in | flat vs. new grad | often $1M+ for strong performers, highly variable | Almost entirely bonus; widest spread of any row |
The Levels.fyi figures for software and quant-dev roles cluster in the mid-to-high $300,000s total, and the per-level ordering on those pages is mostly noise from a small sample. Don’t read much into ‘L1 pays more than L4’ artifacts; at a firm this private with this few data points, the bands blur together and the bonus swamps any banding anyway.
Trader, researcher, and quant dev sit on different curves
The three tracks start close and diverge hard. A trader’s bonus is the one most directly tied to PnL, so the ceiling is highest and the year-to-year swing is roughest. A quant researcher builds the signals and models that desks trade; the pay is excellent and a bit more insulated from any single desk blowing up or printing. A quant developer (Jane Street’s flavor of software engineer, heavy on OCaml and low-latency systems) gets paid like a top-tier engineer and then some, but the bonus is less directly attached to a trading book, so the upside curve is flatter and the floor is steadier.
If you want the highest expected value and can stomach the variance, trading wins. If you want most of the money with less of the whiplash, the dev and research tracks are the trade. People sometimes take the trading seat for the headline number and then discover that a flat year on a flat desk is a real outcome, not a hypothetical.
Why the salary numbers you find online mislead in predictable ways
Three things bend the public figures upward. The people who post comp on Levels.fyi and Reddit are disproportionately the ones who did well and want to say so; nobody screenshots a disappointing number. Year-one totals get puffed up by a signing bonus and, often, a first-year guarantee, so that $600,000 graduate figure is partly one-time money that won’t repeat. And a chunk of the bonus is usually deferred, vesting over a few years, which means the headline ‘total comp’ is not all cash you can spend this April. Prop shops structure it that way on purpose, to keep you from walking to a competitor with the firm’s recent profits in your pocket. The mechanics are the same across the industry; see our writeup on deferred compensation on Wall Street for how the multi-year vesting works.
None of this makes the numbers fake. Jane Street really does pay its new grads more than almost anywhere on earth, and a good trader really can clear seven figures young. The point is to read the figure correctly. A year-one $500,000 is a base plus a one-time guarantee plus a partly-deferred bonus, not a salary you can count on repeating untouched.
How to read an offer when it lands
When the number comes, split it into the part that is contractual and the part that is hope. Base and signing are contractual. A first-year guarantee, if you get one, is contractual for that year only. The discretionary bonus is a forecast, and the recruiter’s ‘people at your level typically see X’ is a forecast with a marketing department attached. Ask what fraction of the bonus is deferred, over how many years, and what happens to the unvested part if you leave. The answers tell you how much of the headline is money you control versus money the firm is holding to keep you around.
Know your own tolerance for variance before you sign, too. The difference between a Jane Street offer and a FAANG offer isn’t only the size of the number, it’s the shape of the distribution behind it. A staff engineer at a big tech company has a comp figure that mostly shows up as promised. A Jane Street trader has an expected value that runs much higher and a variance that can make two adjacent years feel like two different jobs.