Quant trader, quant researcher or quant developer: the work, the pay and the interviews

Students applying to trading firms usually pick a role by its title, and the titles are a poor guide. At one firm a "quantitative developer" writes trading strategies; at another the same title maintains a risk report. Hudson River Trading calls its researchers "algorithm developers". I have worked as a quant developer and as a quant researcher, and the cleanest way I know to tell the three jobs apart is to ask what each person is blamed for when something goes wrong.

What each one is answerable for

A quant trader is answerable for positions. The trader decides, or sets the rules that decide, what the firm holds and at what price, and owns the profit and loss that follows. At a market maker that means setting the parameters of the quoting strategies, watching them through the trading day, and stepping in when a market moves in a way the model was not built for. The feedback is fast: you find out within minutes, sometimes seconds, whether a decision was right.

A quant researcher is answerable for whether a signal is real. The work is to find something in data that predicts prices, volumes or fills, build a model on it, and then try hard to prove it wrong: test it on data it has never seen, account for costs and delays, check that it does not quietly depend on one week in 2020. Most ideas fail that test. The feedback is slow, and a good month can produce nothing anyone outside the team would notice.

A quant developer is answerable for the system working. The developer builds what the strategies run on: market data handlers, order gateways, the simulator researchers test against, the tools traders watch. At a high-frequency firm a large part of the job is latency, keeping the time from a price arriving to an order leaving down to microseconds. A developer's mistakes show up as an outage or a wrong order, which is why the interviews for this role care so much about correctness.

The lines blur in practice. Traders at most of these firms write code every day. Researchers at smaller firms often put their own strategies into production. Many firms have a "research engineer" or "trading engineer" title that sits between research and development. Read the advert for what the person will be answerable for, not for the title.

What the firms pay each role

US pay-transparency laws make trading firms state a base salary in their adverts. Read side by side, one firm's adverts for the three roles, graduate ones where it has them, show how it values each. These are the figures as each firm prints them, read from their own job boards between 5 and 13 October 2026.

Two things stand out. At the firms that pay most, the graduate base is the same for all three roles: HRT and Five Rings advertise $300,000 whichever door you come in through, and Jump and Akuna state the same range or minimum for both. At IMC, DRW, Old Mission and Tower, research and trading sit above development, by $50,000 at the top of the stated range.

The bigger gap is the one the adverts leave out. Every one of them adds a discretionary bonus, and a bonus follows the profit a person can be credited with. That is easiest to measure for a trader, harder for a researcher, and hardest for a developer. Over a career, that is where the roles diverge, and no public advert tells you by how much. Anyone who quotes you a precise figure is guessing.

How each role is interviewed

The three processes share a first filter, a numerical or coding test online, and then test different things.

Traders are tested on speed and judgement under uncertainty. Mental arithmetic is timed: candidates report Optiver's numerical test as 80 questions in eight minutes, scored plus one for a right answer and minus one for a wrong one, and Flow Traders' as 75 questions in ten minutes without a calculator. Then come probability questions, betting and market-making games where you quote a price and get traded against, and a lot of "what would you do now?". Jane Street says plainly that it does not test finance. DRW asks trading candidates to be ready to talk about recent markets.

Researchers are tested on depth. The probability goes further, into conditional expectation, random walks and the distributions behind them, and statistics comes in: regression, overfitting, what a p-value does and does not tell you. Two Sigma lists its research interview topics as open-ended data problems, coding and algorithms, and statistics or the candidate's research field. G-Research describes typically four one-hour technical interviews, one of them in-depth maths. Citadel's research process opens with a coding interview on CoderPad and ends with three to five hour-long interviews on site. At Jane Street, a quant researcher's interviews are a mix of the trading interviews and the software engineering ones.

Developers are tested on code that is correct. HRT describes its engineering process as phone screens and then on-site interviews in coding, debugging, design and fit, in C++ or Python, and says it avoids questions that need a single burst of insight. DRW sends developers a Codility test before the interviews. At the low-latency end, expect questions on how memory, caches and threads actually behave, because that is the job: we wrote up twelve of those questions separately.

Which one suits you

Background matters less than people think, and temperament more. A few honest signs:

• If you are happy to be wrong in front of others several times a day, as long as you are right more often, and you enjoy games of incomplete information, look at trading. Many good traders studied maths, physics or engineering, not finance.
• If you would rather spend three weeks proving an idea does not work than three minutes guessing whether it does, look at research. Firms hire researchers at bachelor's, master's and PhD level, but the PhD tracks are separate. Two Sigma's research internship asks for a research project on real data.
• If what satisfies you is a system that runs for months without failing, and you care how fast it is, look at development. A computer science degree is the usual route, and strong C++ opens the high-frequency firms.

You do not have to choose for ever. Developers who learn the research side move across often, and firms with combined roles, such as HRT's algorithm developers, expect people to do both. The reverse move is rarer, mostly because good engineering takes years of habits that a researcher has had no reason to form.

PDF · Edition 2 · £79

The Quant Careers Report

Pay, eligibility dates, application windows and interview processes at 13 trading firms, role by role, with 345 practice questions written for the report. Every figure sourced.

Buy the report — £79 See what is inside

Questions people ask

The Quant Careers Report Practice questions

Practice for each route: timed mental maths and market-making questions for traders; probability and statistics and machine learning for researchers; the coding course and the low-latency C++ course for developers.