G-Research and XTX Markets are the two London firms most often named together by machine-learning students, but for a 2027 graduate they are not the same kind of door. G-Research is the one that is open: it had 67 London vacancies on 14 October 2026, eight of them internships or graduate roles, with fixed 2027 dates. XTX has the higher firm-level ceiling and the smaller team, but its job board listed seven roles that day and none for students, so most graduates apply to G-Research on a timetable and to XTX opportunistically.
The rest of this post is the detail behind that: how each interview loop is built, what the filed accounts actually say about pay, and which profile should push harder at which firm.
The 30-second verdict
| G-Research | XTX Markets | |
|---|---|---|
| What it is | London quant research and technology firm; ML and statistics to forecast markets | Algorithmic market maker; ML price forecasts, principal capital only |
| Size | 800+ staff (Quantt estimate, 2026) | Just over 300 (Young and Calculated, Aug 2026); ~250 per Quantt (May 2026) |
| Open roles, 14 Oct 2026 | 67, all London; 8 student or graduate roles | 7 (London, Singapore, Finland); 0 student roles |
| Hiring volume | Not published | ~30-50 researchers and engineers a year (Quantt, May 2026) |
| First gate | 10-question, 90-minute multiple-choice quiz | 90-120 minute OA; ~5-10% pass (Quantt; Tradermath) |
| Rounds after the gate | Triage, four one-hour technicals, leadership | 60-minute phone screen, then 5-6 interviews in one day |
| QR student bar | Master's or PhD, final or penultimate year | XTY Labs AI internship (NY) was PhD-only when last open |
| Public pay data | Two staff LLPs: £163.5k and £2.17m averages (yr to Mar 2025) | £457k average, 127 staff, one entity (2025) |
| Firm results | Not published | £3.93bn revenue, £1.71bn profit after tax, 2025 |
| Visa | Relocation and immigration costs covered | "Visa & Immigration Support" listed as a perk |
Interviews: a quiz gate vs an OA gate
Both loops cut most people before a human hears them talk. The gates test different things.
G-Research sends researchers a quiz of 10 multiple-choice questions with five options each and 90 minutes on the clock (Techinterview, August 2026). Tradermath (September 2026) lists the syllabus as probability, statistics including OLS, linear algebra, calculus with differential equations, programming and finance, and says there is a separate ML-specific version for ML roles. After that comes a triage interview, four hour-long interviews, and leadership conversations. The hour Tradermath singles out is mathematics: candidates fail it because it goes well beyond quiz level. G-Research's own students page puts the quant-research internship process at roughly one to two weeks: application, an online aptitude test, technical interviews, and a final interview with a functional lead.
XTX sends an online assessment after a CV screen: for engineers, two or three hard algorithmic problems in 90-120 minutes; for researchers, a combined ML and probability test of similar length (Tradermath, July 2026). Then a 60-minute phone screen where, per the same guide, interviewers often push back on correct answers, then a super day of five to six back-to-back interviews. Tradermath's candidate-reported mix for researchers is about 40% machine learning, 25% probability and statistics, 25% coding and 10% behavioural. Sample questions it lists: derive E[X | X+Y=z] for independent standard normals; keep a rolling standard deviation over the last N points in O(1) amortised time; explain why K-fold cross-validation leaks in time series. Application to offer typically takes four to six weeks, and rejected candidates wait 12 months to reapply.
The practical difference: G-Research's quiz rewards breadth across a fixed syllabus under time pressure. XTX's loop goes deeper on ML validation and on production-grade coding, and has more rounds.
Culture and pay
Neither firm publishes a graduate salary, so be careful with every number below.
XTX. eFinancialCareers (April 2026) reported that XTX Markets Technologies Limited, one of its UK entities, employed 127 people in 2025 at an average of £457k, up about 5% from £436k in 2024. The firm itself made £3.93bn net revenue and £1.71bn profit after tax across three UK entities in 2025, up 43% and 33% (Finance Magnates, April 2026). Young and Calculated (August 2026) put net revenue per employee at about £13.1m and the research cluster at more than 25,000 GPUs. On levels.fyi (updated 14 October 2026) the overall XTX median is $384,956, but the sample is small enough that its software-engineer median sits at only $141,788, so it says little about a London offer.
G-Research. Its pay is split across staff LLPs. eFinancialCareers (January 2026) reported Alastair LLP's 257 staff averaged £163.5k and Braunford LLP's 49 averaged £2.17m in the period to March 2025; Tradermath (September 2026) labels the first group administrative and the second non-administrative. Tradermath's London-wide estimate for a first-year quant researcher is £160k-310k base plus bonus, which is a market range, not a G-Research figure. levels.fyi (updated 14 October 2026) shows a G-Research software-engineer median of $250,746. Its posted benefits are concrete: 35 days' leave for office-based staff, 9% pension, private health cover, and accommodation for interns.
Culture, from the community. Most public threads are old, so weigh them lightly. In a June 2023 Blind thread, a Meta SRE in London weighing Citadel against XTX got replies calling XTX's culture and work-life balance good, with one commenter adding that a firm that small is likely more selective. In a 2021 Blind thread, a London engineer with a G-Research offer described it as a prop shop that is hard to understand from the outside; the one reply told them to finish their Jane Street process before signing. That opacity is still real: G-Research publishes no revenue, and eFinancialCareers reported it cut about 100 London technology staff in 2024.
One recent signal of each firm's priorities: G-Research's news page is running engineering explainers (Apache Arrow on 5 October 2026, Parquet on 21 September) and monthly research-grant awards. XTX on 16 September 2026 committed another $10mn each to Lean FRO and the Mathlib Initiative, on top of more than $50mn already given to AI-for-maths projects.
Which should you choose?
- Undergraduate in CS or ML: G-Research. The Graduate Machine Learning Engineer role (start 31 August 2027) accepts current undergraduates, per its Built In London listing, and the engineering internships run 12 or 24 weeks. Watch XTX's board for engineering roles; its Singapore C++ engineer posting says most of the existing team is self-taught and a CS degree is not necessarily required.
- Master's student aiming at research: G-Research's Quant Research Internship (21 June to 27 August 2027) is built for you, PhD preferred. XTX has no matching open role today.
- ML PhD with publications: apply to both. G-Research has dedicated ML research and NLP internships; XTX's XTY Labs AI internship in New York paid $35,000 a month for 12-14 weeks when the Telegraph reported it (September 2025), but it is not on the board now.
- You want the smallest possible team and the most compute per head: XTX, if a seat opens.
- You need a predictable timeline for a visa plan: G-Research, because its dates are fixed and published.
For international students: both firms say they support immigration, but the UK clock matters more than the employer. The Graduate visa lasts two years if you apply on or before 31 December 2026 and 18 months from 1 January 2027; PhD graduates get three years (GOV.UK, retrieved October 2026). A Master's student graduating in late 2026 who applies before the cutoff keeps the longer window for an XTX opening that may not appear on a campus timetable.
Prepare for both
- Track every 2027 London quant deadline on the New Grad 2027 tracker.
- Read the full guides: how to get into G-Research and how to get into XTX Markets, plus which UK quant firms sponsor visas.
- Both loops push on statistics and ML reasoning. The Two Sigma quantitative researcher mock is the closest format; Applr's mock follows up on why your answer holds, which is what XTX's phone screen tests.
Sources
- G-Research, Vacancies, retrieved 14 Oct 2026
- G-Research, "Quant Research Internship" vacancy, retrieved 14 Oct 2026
- G-Research, Graduates and internships page, retrieved Oct 2026
- G-Research, News, retrieved 14 Oct 2026
- Built In London, G-Research "Graduate Machine Learning Engineer" listing, retrieved Oct 2026
- XTX Markets, Greenhouse job board, retrieved 14 Oct 2026
- XTX Markets, Careers, retrieved Oct 2026
- XTX Markets, "C++ Software Engineer" posting (Singapore), retrieved 14 Oct 2026
- XTX Markets, "Update on XTX Markets' AI for Maths Philanthropy", 16 Sep 2026
- Techinterview, "The G-Research interview and the 90-minute quiz that gates it", Aug 2026
- Tradermath, "G-Research Interview Guide (2026)", Sep 2026
- Tradermath, G-Research salary page, updated 28 Sep 2026
- Tradermath, "XTX Markets Interview Guide", 31 Jul 2026
- Quantt, "XTX Markets Interview", 9 May 2026
- Quantt, G-Research firm profile, 2026
- eFinancialCareers, "Pay at XTX Markets rose modestly", 7 Apr 2026
- eFinancialCareers, "London's most secretive electronic trading firm is obfuscating its pay", Jan 2026
- Finance Magnates, "XTX Markets Revenue Rises 43% to £3.93 Billion in 2025", 3 Apr 2026
- Young and Calculated, "XTX Markets: The 300-Person Firm Trading 40x the London Stock Exchange", 2 Aug 2026
- levels.fyi, XTX Markets salaries, updated 14 Oct 2026
- levels.fyi, G-Research salaries, updated 14 Oct 2026
- The Telegraph via Yahoo Finance, "British billionaire trader to pay interns £26,000 a month", 4 Sep 2025
- Blind, "FAANG → Citadel/XTX move thoughts", 11 Jun 2023
- Blind, "G-Research in London", Sep 2021
- GOV.UK, Graduate visa, retrieved Oct 2026
Frequently asked questions
- Is XTX Markets harder to get into than G-Research?
- For a graduate, yes, mostly because XTX has so few seats. Quantt (May 2026) estimates XTX hires roughly 30 to 50 researchers and engineers a year worldwide, with an online-assessment pass rate of about 5 to 10%, and on 14 October 2026 its Greenhouse board listed seven roles and no graduate or intern posting. G-Research's own vacancy page listed 67 London roles that day, eight of them internships or graduate roles.
- Which pays more, G-Research or XTX?
- Neither publishes graduate pay, and the public numbers are averages that mix juniors and seniors. eFinancialCareers (April 2026) reported XTX Markets Technologies Limited paid an average of £457k across 127 staff in 2025. For G-Research, eFinancialCareers (January 2026) reported Alastair LLP's 257 staff averaged £163.5k and Braunford LLP's 49 averaged £2.17m in the period to March 2025. Neither figure is a graduate offer.
- Can undergraduates apply to G-Research or XTX?
- At G-Research, yes, but not to quant research: the 2027 Quant Research Internship asks for final- or penultimate-year Master's or PhD students, while the Graduate Machine Learning Engineer role starting 31 August 2027 accepts current undergraduates, per G-Research and Built In London listings retrieved October 2026. XTX has no student posting open as of 14 October 2026; its engineering postings judge skill rather than degree.
- What is the difference between the G-Research and XTX interviews?
- G-Research gates on a 10-question, 90-minute multiple-choice quiz (Techinterview, August 2026), then runs a triage call, four hour-long interviews including a deep mathematics hour, and leadership interviews (Tradermath, September 2026). XTX starts with a 90-120 minute online assessment, then a 60-minute technical phone screen, then a super day of five to six 45-60 minute interviews in London (Tradermath, July 2026).
- Do G-Research and XTX sponsor visas?
- Both say they support immigration. G-Research's internship posting says it covers relocation and immigration costs and provides an immigration consultant; XTX's careers page lists visa and immigration support as a perk. Separately, the UK Graduate visa lasts two years if you apply by 31 December 2026 and 18 months from 1 January 2027, with three years for PhD holders (GOV.UK, retrieved October 2026).