G-Research Interview Questions & Process (2026 Guide)

What the online quant quiz, technical interviews, and leadership round actually test — based on candidate reports and the firm's own guidance.

Updated July 3, 2026 · QuantVault

G-Research is a London-based quantitative research and technology firm that uses machine learning and statistical methods to predict movements in financial markets. It is consistently described as one of the highest-paying quant employers in the UK, and it hires heavily from maths, physics, and CS PhD programmes. Most candidates first encounter the firm through its public puzzle competitions — we cover those separately in our G-Research weekly quiz guide — but the actual hiring process is a different, longer pipeline.

The reported process at a glance

G-Research's own career guidance describes a four-phase process for quantitative researcher roles. Candidate reports on Glassdoor and QuantNet are broadly consistent with it. As of mid-2026, the commonly described format is:

Stage	Format	What it tests
1. Application	CV review, typically ~1 week	Research background, maths/CS pedigree
2. Online quant quiz	Candidates report 10 multiple-choice questions, 5 options each, 90 minutes	Probability, statistics (incl. OLS), linear algebra, calculus/ODEs, some programming and finance
3. Technical interviews	General track: four one-hour interviews; ML track: two one-hour interviews	One deep mathematics round; others drawn from stats, ML, programming, finance
4. Leadership interviews	Conversations with senior staff	Fit, motivation, research judgment

Glassdoor data puts the average time-to-hire for quantitative researcher candidates at around 19 days, with an interview difficulty rating of roughly 3.6 out of 5. That is demanding but rated below the very hardest names on our quant firms ranked by interview difficulty list — the filter here is breadth, not trick puzzles.

Stage one: the online quant quiz (the "maths test")

This is what most people mean when they search for the G-Research maths test. According to the firm's guidance, you sit one of two versions depending on your background: a general quantitative aptitude assessment or an ML-specific quiz. Candidates consistently report the general version as 10 multiple-choice questions in 90 minutes — so nine minutes per question, which tells you these are not mental-math sprints. Reported topics include:

Probability and stochastic processes — candidates have reported variants of classic setups like gambler's ruin; our probability question bank covers this territory.
Statistics, especially regression — OLS comes up by name in multiple reports, including questions about minimising a cost function and how the solution scales with initial conditions.
Linear algebra and calculus — differential equations get specific mention, which is unusual among quant OAs.
Light programming and finance — present but secondary.

To calibrate the level, here is a representative practice problem in the same OLS style (this is ours, not a leaked G-Research question). For regression through the origin, minimise $\sum_i (y_i - \beta x_i)^2$ over $\beta$. Setting the derivative to zero gives

$$\hat\beta = \frac{\sum_i x_i y_i}{\sum_i x_i^2},$$

and you should be able to say immediately why this differs from the standard OLS slope (no demeaning) and what happens to the variance of $\hat\beta$ as the $x_i$ spread grows. That's the register of the quiz: single clean derivations, done accurately under time pressure.

Stages two and three: technical and leadership interviews

For the general quant track, G-Research describes four one-hour technical interviews, one of which is an in-depth mathematics round; candidate reports say the remaining interviews are drawn from statistics, finance, programming, and machine learning. For the ML track, the firm describes two one-hour interviews mixing ML knowledge with maths, programming, and statistics. Interviewers reportedly do not expect you to know everything — the questions are described as hard but doable, with an emphasis on reasoning aloud.

The statistics and ML rounds sit closer to a research seminar than a brainteaser session: expect to defend assumptions, discuss overfitting and validation, and work through estimator properties. Our statistics bank and machine learning questions map directly onto this. Successful candidates then meet company leadership for fit and motivation conversations.

How to prepare

G-Research's own guidance recommends 150 Most Frequently Asked Questions on Quant Interviews for the general track, and Cracking the Coding Interview plus The Elements of Statistical Learning for ML candidates, alongside practice on Project Euler and Kaggle. A sensible plan mapped to that:

Weeks 1–2: rebuild probability and expectation fundamentals, then drill linear algebra and calculus/ODE mechanics for the quiz.
Weeks 3–4: go deep on regression and inference — OLS assumptions, bias–variance, hypothesis testing — since statistics is the reported centre of gravity.
Final week: timed mixed sets under quiz conditions, plus coding questions at a standard-DSA level and mock interviews where you narrate your reasoning.

One honest caveat: G-Research signs candidates to fairly strict confidentiality, so post-2024 question-level reports are thinner than for firms like Optiver or Jane Street. The structure above is well-corroborated; specific question content varies, so prepare for breadth rather than memorising leaks.

Ready to practise? Work through our probability and statistics banks under time pressure, then browse the full problem library — 2,800+ worked problems, around 400 of them free.

More firm guides

Frequently asked questions

What is the G-Research online assessment format?

Candidates consistently report a 10-question multiple-choice quiz with 5 options per question and a 90-minute limit. G-Research says you sit either a general quantitative aptitude version or an ML-specific version depending on your background. Reported topics include probability, statistics (especially OLS regression), linear algebra, calculus and differential equations, plus some programming and finance.

How many interview rounds does G-Research have?

The firm describes four phases: application review, an online quant quiz, technical interviews, and leadership interviews. For the general quant track the technical stage is four one-hour interviews including one deep mathematics round; the ML track is described as two one-hour interviews mixing ML, maths, programming, and statistics.

How hard is the G-Research maths test?

Glassdoor respondents rate the overall quant researcher process around 3.6 out of 5 for difficulty. The quiz allows roughly nine minutes per question, so it rewards careful derivation rather than speed arithmetic. Candidates describe the questions as fairly standard for quant research interviews, with a heavier tilt toward statistics than at market-making firms.

Is the G-Research weekly quiz the same as the interview test?

No. The public weekly/monthly puzzle quiz is a marketing and community competition, while the hiring assessment is a private timed multiple-choice test sent after CV screening. The public quiz is still useful practice for the style of mathematical thinking the firm likes, but doing well on it is not part of the formal process.

Practice the real thing

QuantVault has 2,800+ quant interview problems with full solutions, intuition, and hints, firm-by-firm interview funnels, and an auto-graded coding judge. Start free.

Browse problems Firm interview guides