Marshall Wace is a London hedge fund founded in 1997 by Paul Marshall and Ian Wace. It runs long/short equity strategies, fundamental and systematic, and its TOPS platform, launched in 2002, is described by the firm as the world's first alpha capture application. KKR has held a minority stake since 2015. The firm's own postings put it at over $70bn in assets with 700 to 900 staff across London, New York, Hong Kong, Singapore, Shanghai and Abu Dhabi.
Marshall Wace is unusual in how much it tells you. It publishes detailed application guides with sample questions, requires an in-person final stage, asks you to apply to one role and location only, and prohibits candidates from seeking out the content of its tests and interviews. This guide therefore describes formats and question families, not leaked questions.
There are 2 different processes. Technology interns and graduates start with a Codility coding test, which 3 candidates who sat it in late 2025 and 2026 say allowed AI tools. Quant research interns for 2027 start with a new in-person, invigilated Pearson exam of 18 questions with negative marking, which replaces the 3 online tests used last cycle. Neither track has a numerical reasoning test or a game-based assessment.
- 1Online or in-person test. Codility for technology. A Pearson test centre exam for quant research.
- 2Screening call. 15 to 30 minutes with recruitment or Human Capital on motivation and basics.
- 3Lab or project stage. Technology: a lab test on a virtual machine. Quant: a presentation of a project you built.
- 4Technical interview. A 1-hour problem that grows in complexity, with 2 interviewers.
- 5On-site day. Up to 4 sessions for technology. A programming test, a brainteaser interview and manager interviews for quant research.
The 2027 programmes
| Programme | Offices | Dates | Who |
|---|---|---|---|
| Technology Intern | London, New York, Singapore, Hong Kong | London 28 June to 3 September 2027. New York 14 June to 22 August. Singapore 17 May to 25 July | Students not graduating in 2027. First-years were eligible in 2025 |
| Technology Graduate Programme | London, others by year | 18 to 24 months, rotations across up to 6 teams | Final-year students and graduates |
| Quant Research Intern | London only | 28 June to 3 September 2027 | Penultimate year of a STEM master's or PhD, finishing in 2028 |
| Quant Associate Programme | London, New York | Full time, research and implementation tracks | Master's or PhD, or 2 to 3 years of experience |
| AI Placement and Technology Placement Year | London | April to September 2027 | Imperial students and placement students |
Applications for the quant research internship opened on 17 August 2026, with tests from 21 September, and the firm expects to fill the positions by December. Technology applications are rolling. Past technology intern projects include a backtesting framework for the systematic trading system, distributed coding agents and a user analytics platform, and interns work in pairs under a senior engineer.
The technology track
Stage 1: Codility test
| Item | What the firm and candidates say |
|---|---|
| Invite | Sent directly by Codility, to complete within 7 days |
| Length | About 70 minutes for most candidates, 60 for one |
| Tasks | Candidates disagree: 3 tasks (1 coding, 1 SQL, 1 debugging) or 11 questions. The test may include behavioural multiple choice |
| Topics | The firm lists general coding aptitude, error handling and sequence processing |
| Languages | 18 are listed, but some tasks fix the language. Elixir has come up, and the languages on the invitation may not match the ones the task allows, so check before you start |
| AI tools | Reported as allowed by 3 candidates between November 2025 and March 2026, who describe being encouraged to use LLM tools. The firm publishes no rule for the technology test |
| Retake | None in the same season, including after a failed test or screening call. Reapplying next season is welcome |
Reported problem families are data handling rather than algorithms: parsing JSON event data, counting the maximum number of events in any 60-second window from sorted timestamps, a SQL query for users with successful logins from 2 or more countries, and debugging code in an unfamiliar language.
Read the invitation and each task header for the language and the AI rule before you start. The rules appear to be set per invitation, and the one official AI rule Marshall Wace has published is the opposite one, a ban on LLMs in last cycle's quant Codility test. If you have sat the test recently, tell us what you saw.
How to prepare
- Data handling in Python. JSON parsing, sliding windows over timestamps and deduplication. The parsing lessons in the developer course cover the shape.
- SQL. GROUP BY with HAVING and COUNT(DISTINCT), written by hand against a small table until the syntax costs nothing.
- Reading code you do not know. Practise fixing a small bug in a functional language under 15 minutes. The fix-the-bug lesson teaches the method.
Stage 2: Screening call
15 to 30 minutes with recruitment. Reported questions: explain what an API is to a 10-year-old, explain binary search to a 3-year-old, explain a CI/CD pipeline, explain a concept from your CV to a non-technical person, and why finance and technology. The application form also asks which technology outside AI and machine learning has caught your attention, and the firm says that answer comes back up in a later interview.
Candidates who did not progress describe a recruiter reading questions fast from a tablet and being unimpressed by vague answers. Prepare 3 plain-language explanations in advance and have a real answer to the technology question.
Stage 3: Lab test
About 60 minutes alone on a virtual machine, then a 20 to 30 minute debrief with 1 or 2 graduates. The firm describes technical and non-technical questions, some asking you to review code, others to support queries that have come in. Candidates report about 15 straightforward questions: what is wrong with a piece of JavaScript, what the S in HTTPS stands for, what shorting is, code reviews, and writing the comments of a pull request and saying whether you would accept it. No official AI rule is published for the lab, and the current guide's summary says resources must stay inside the supplied environment, so assume no outside tools.
How to prepare
Review a pull request and write the comments out loud. Debug a small API. Have one-sentence answers for basic web and finance terms. The code reading lesson is the right preparation for the review half.
Stage 4: Virtual technical interview
1 hour with 2 members of the technology team. One technical question that develops in complexity, on data structures, algorithms, data processing and complexity analysis, usually without a live programming environment. Expect questions on efficiency, improvements and alternatives, plus your CV.
Stage 5: On-site
Up to 4 sessions, and the firm says it may cancel the later ones if an earlier one goes badly. The technical round comes first with 1 or 2 programming questions, then a senior interview on competency and culture, then Human Capital, and for interns a presentation of a project you have written. The firm insists on real code: it warns that research projects are often too theoretical, asks group-project candidates to show their own contribution, and wants you to explain every library you used, especially any machine learning component. Reported questions include why you used a WebSocket rather than server-sent events or REST, and have a personal project to talk about: candidates without one have been rejected.
How to prepare
Pick a personal project with code you can defend line by line and rehearse the library choices and their alternatives. The project defence lesson covers how interviewers probe it. The Quant Dev 50 is the right set for the on-site programming questions.
The quant research track
Stage 1: Pearson exam
| Item | What the firm says |
|---|---|
| Format | A computer-based exam at a Pearson test centre with a live invigilator. The invite comes directly from Pearson |
| Questions | 18, drawn from probability, statistics, regression, algebra, matrices, optimisation and algorithm design |
| Scoring | 2 marks for a correct answer, minus 1 for a wrong one, 0 for a blank. No score breakdown is given |
| Length | 2 hours in the Quant Associate guide. The intern length is not stated separately |
| Materials | A reusable notepad and marker. No personal items. The calculator rule is not stated |
| Booking | One rebooking with 48 hours of notice on the Pearson page, 72 hours in the associate guide. A no-show closes your application |
The firm's own sample questions set the level: the expected mean IQ of a sample of 25 students when the first is 150, the average complexity of quicksort, every solution of an equation of the form (a quadratic in x) to the power of (another quadratic in x) equal to 1, and a digit puzzle with 8 distinct digits. Candidates who sat the Correlation One test this replaces describe 20 questions on probability, statistics, regression, linear algebra and calculus with a few brainteasers, such as counting robot paths on a 5 by 5 grid with a blocked square, the variance of a uniform variable, the trace of a matrix exponential, and the probability that 51 coins show strictly more heads than 50 coins.
No candidate had reported the Pearson exam when this guide was written. Candidates from last cycle describe a 60-minute Codility test, a 60-minute Correlation One technical test and a 45 to 50 minute data test on CSV files, each with a 7-day window and sequential gating. Those are the old format.
How to prepare
Under 2 marks for a right answer and minus 1 for a wrong one, a guess pays only if you are more than one third sure. Work the sample questions, then counting, OLS, covariance matrices and the optimisation puzzles. The probability questions cover the coin and grid problems.
Stages 2 to 4: Screen, presentation and on-site
A 30-minute Human Capital screen on Zoom covers motivation and competencies. The technical project presentation is around 15 to 20 minutes with the rest for questions, and a university assignment, thesis or personal project is fine. The on-site day has 3 parts.
- Programming test. On a laptop, you review written code and answer questions on core programming principles. The firm's sample is the minimum-cost staircase problem.
- Brainteaser interview. 50 minutes, 2 brainteasers. The firm's sample is a dice game where players cross out numbers as they first appear and the player who crosses out the last one wins: would you play first or second? Candidates also report counting the painted unit cubes on a 10 by 10 by 10 cube and finding the smallest positive integer whose digits multiply to 10,000.
- Manager interviews. Informal conversations with 2 or 3 managers at a time, plus shadowing a quant.
The firm says it does not expect much extra preparation for the technical interviews and that it has a very large bank of questions. The Reasoning questions and the game theory course cover the brainteaser round.
Key tips for success
- Apply to one role and location. The firm says multiple applications delay review.
- For technology, check the language and AI rules on the invitation itself and do not assume the Codility rule carries into the lab.
- Have a personal project with real code. It is a filter at the presentation stage.
- For quant research, book the Pearson exam early and bring identification. A missed booking closes the application.
- Expect a long process. Candidates describe 7 rounds over about 3 months for the technology graduate scheme.
Closing remarks
Marshall Wace tests engineers on real engineering: data handling, debugging, code review and defending code they wrote, with AI reported as allowed on the first screen and nowhere after it. It tests quant researchers on mathematics under exam conditions, then on a project and 2 brainteasers in person. The formats changed for 2027, so weigh any candidate account by its date.
Prepare for the stage in front of you, keep your application confined to one role, and treat the firm's own guides as the primary source. Work through everything we have for Marshall Wace, and read our quant developer interview guide or our quant researcher interview process guide for how these stages compare across firms.