The First Cut Is the Deepest
Eight minutes, 80 questions. That is the reported format of Optiver's mental arithmetic test, and for thousands of applicants each autumn it has been the entire interview. Across the industry, the online assessment is where most quant applications end, usually within two weeks of applying.
This is odd, because candidates prepare backwards. They spend months on interview brain teasers they will never be asked, because the OA filtered them out in October. The OA deserves preparation in proportion to its position in the funnel: first, brutal, and largely standardised.
This guide covers what quant OAs actually test, then the reported format at each major firm - Optiver, IMC, SIG, DRW, Citadel, Jane Street, Virtu, Akuna, Two Sigma and HRT - and how to prepare for each style. One caution up front: firms change formats without notice, and everything below is drawn from candidate reports rather than official specifications. Treat it as a map from last season.
What Quant OAs Test
Nearly every quant OA is built from four ingredients, mixed differently by track.
Mental arithmetic under time pressure. Two-digit multiplication, fractions, decimals and percentages at a pace that punishes hesitation. Trading firms use this as a proxy for comfort with numbers when it matters. Speed here is trainable, which is exactly why firms test it; our 80 in 8 simulator and mental maths questions drill this specific skill.
Probability and expected value. Dice, cards, coins, and increasingly short game scenarios: should you take this bet, what is the fair price of this payout. These test whether you reason about uncertainty natively or by formula.
Pattern recognition and logic. Number sequences, matrix puzzles (Raven's-style), and abstract reasoning items, often adaptive. Common in the Dutch market makers' batteries. Less trainable than arithmetic, but familiarity with the question grammar is worth real marks.
Coding. For developer, researcher and increasingly trader tracks: one to three algorithmic problems on HackerRank, CodeSignal or Codility, typically LeetCode medium to hard. Edge cases and efficiency both count, and partial credit rarely rescues a slow solution.
Did you know?
Tap a tile for a quick interview snippet — then keep going.
Optiver
The most famous OA in the industry. For years Optiver's trading assessment led with the "80 in 8": 80 arithmetic questions in 8 minutes, reported as multiple choice with four options, with wrong answers penalised (commonly minus one mark against plus one). Optiver publishes no pass mark; the figure candidates and prep sites repeat most is about 55 net, with 70 or more called competitive.
The rest of the battery, as candidates describe it, covers number sequences (NumberLogic), probability (Beat the Odds), short cognitive games (Zap-N) and at times a personality questionnaire. The 2026/27 invites candidates have quoted add three new modules (Likelihood-list, Intervals and Orderbooks) and do not list the 80 in 8, so check your own invite. Optiver's Europe FAQ says the assessments can only be taken once every eight months. Our Optiver 80 in 8 guide covers the evidence in detail, and full process details are in our Optiver interview guide. Practise the pace on our 80 in 8 challenge.
IMC
IMC's trader test changes more than most. US candidates for the Summer 2027 internship describe a 40-minute SHL assessment: a personality questionnaire, then inductive reasoning puzzles built on shape and pattern rules. In 2025, trader candidates in Europe and the US played BrainsFirst's NeurOlympics games instead, and in 2023 US graduates reported 15 maths and probability questions in an hour. Most US candidates describe a CV screen before the test is sent.
Engineering candidates consistently report two HackerRank problems in two hours. For the trader test, find the most recent reports for your region; for the SHL version, untimed practice on pattern rules is the best start. Our IMC interview guide has the details by cycle.
SIG (Susquehanna)
SIG's job ads say the online assessment is run by Mercer | Mettl. Candidates in the 2026 and 2027 cycles describe a Problem Solving Assessment of 17 questions in 60 minutes: probability, expected value, combinatorics, Bayes, Markov chains, a little calculus and logic puzzles. Trader candidates used to get a shorter Quantitative Evaluation of about 20 minutes, and candidates report it being replaced by the longer test this cycle. Software candidates mostly report CodeSignal.
The probability is first-course material, but the questions are built so that the intuitive answer is often wrong. Our SIG interview guide has worked examples in the same style.
DRW
DRW describes its trader test as "a quantitative challenge focused on probability and logic skills", taken at home under timed conditions. Candidates report about six questions in 45 minutes on probability, statistics, Markov chains and linear algebra, with no speed arithmetic. Software developers take a Codility test, which DRW confirms, and candidates report three problems in 150 minutes.
The linear algebra is what catches people out, since few other firms test it. Our DRW interview guide covers the full pipeline from the test to the Superday.
Citadel and Citadel Securities
"Citadel online assessment" almost always means HackerRank. Reports consistently describe two to three algorithmic problems in 60 to 90 minutes for software engineering and quant research applicants, at LeetCode medium-to-hard difficulty, with hidden test cases where partial credit exists but full passes dominate outcomes. Some quant researcher tracks add multiple-choice statistics and probability questions.
The distinguishing feature is the bar: with applicant volumes in the tens of thousands, candidate folklore holds that anything short of all test cases passing is risky. Practise timed pairs of problems, and rehearse the discipline of reading constraints before writing code. See our Citadel interview guide for what follows.
Jane Street
Jane Street is the notable partial exception on this list. For trading roles the firm has historically gone straight to phone interviews rather than gating on a standardised OA, though some cycles and regions have reportedly used short online maths and logic quizzes, and its focus programmes use puzzle-based applications. Software engineering applicants may receive a take-home-style coding assessment in the language of their choice (OCaml not required).
If you receive one, expect probability and estimation in the firm's distinctive style: fewer questions, more depth, marks for reasoning rather than raw speed. The Jane Street interview guide covers the phone rounds that follow.
Virtu
"Virtu HackerRank" is the search term because that is the process: Virtu's OA for developer and quant roles is a HackerRank set, commonly reported as two or three problems in about an hour, mixing algorithms with occasional C++-specific or debugging questions reflecting the firm's low-latency stack. Reports suggest the difficulty is a notch below Citadel's, with more weight on clean, correct code than on exotic algorithms.
Brush up on standard data structures, know your chosen language's standard library cold, and if you claim C++ on your CV, expect it to be tested.
Akuna Capital
Akuna's assessments vary sharply by track. Developer candidates report HackerRank problems with language-specific questions (C++ for core roles, Python for data roles). Trading candidates face maths and probability batteries, and the firm is known for its Options 101 course, a self-paced options-theory module that some trading applicants complete as part of the funnel, with a quiz at the end.
The options material is a gift: it is the rare OA component where the firm hands you the syllabus. Complete it properly rather than skimming.
Two Sigma and HRT
The research-heavy end of the market. Two Sigma's OA is reported as HackerRank problems (commonly two in 60 to 90 minutes) for engineering, with quantitative researcher applicants sometimes receiving statistics and probability multiple choice instead. HRT's assessment reportedly leans hard on algorithms: candidate reports describe problems at the difficult end of the scale, consistent with a firm whose interviews are among the most demanding in the industry.
For both, the preparation is the same as Citadel's HackerRank track, tilted harder: timed hard-level problems, and for HRT specifically, comfort with performance-sensitive thinking rather than just correctness.
General Strategy
When to take it. Most OA links expire in 3 to 14 days. Take it early in the window but never cold: even two or three days of format-specific practice moves scores meaningfully. If you are applying to several firms, sequence them: take a lower-stakes OA first as a live rehearsal for the ones you care about.
How to practise. Arithmetic speed responds to daily drills on Quantt's 80 in 8 simulator (the Optiver pace and +1/-1 penalty, with typed whole-number answers), the mental math trainer, and Zetamac-style timers for extra volume; two to six weeks of 20 minutes a day is the standard prescription. Coding OAs respond to timed problem pairs under contest conditions. In both cases, practise with the penalty structure you will face: if wrong answers cost marks, train the skip reflex.
Logistics. Stable internet, external mouse, silence, and a glass of water. Many OAs are webcam-proctored; read the rules so you do not lose a result to an accidental violation. Paper is usually allowed for maths batteries but check.
Quick quiz
OA strategy check
Mindset that survives the timer.
Best approach when stuck on an OA item?
After the OA
Results timelines vary from instant (some adaptive batteries) to several weeks. A pass typically routes you to a phone or video interview within a fortnight; silence for a month usually, though not always, means no. Most firms allow reapplication after 6 to 12 months, and OA failure rarely blacklists you.
Keep notes immediately after each attempt: formats, timings, question styles you struggled with. Your OA notes from October are the cheapest preparation you will ever have for next season, and the interview rounds that follow draw on the same skills at greater depth; our full quant interview prep guide picks up from here.
Where This Guide Goes Stale
The obligatory caveats, stated plainly. Everything above is assembled from candidate reports, forum accounts and public postings, not from the firms, and OA formats change between cycles without announcement: firms switch vendors, add game-based modules, retune thresholds and A/B test formats by region and track. Reported pass marks are folklore with survivorship bias baked in; people who pass write more posts than people who fail.
So treat the firm sections as strong priors, not specifications. The invariant part is the skill set: fast accurate arithmetic, native expected-value reasoning, pattern fluency and timed coding. Prepare those and format changes become cosmetic.
Compensation & recruiting notes
Assessment formats, thresholds and timelines described here are reported by candidates and public sources, not confirmed by employers, and change frequently. Nothing in this guide guarantees an assessment format, an interview or any recruiting outcome. Check each firm's official careers communications for current details.
Frequently Asked Questions
What is a quant online assessment?
The first screening stage of most quant applications: a timed online test covering some mix of mental arithmetic, probability, pattern recognition and coding, depending on the role. It typically arrives within days of applying and eliminates the majority of the applicant pool.
How hard is the Optiver 80 in 8?
Hard, by candidate reports: 80 multiple-choice arithmetic questions in 8 minutes with penalties for wrong answers, and a reported pass line around 55 net that Optiver has never confirmed. It rewards several weeks of daily timed practice more than raw talent. Our 80 in 8 challenge mirrors that pace, and our 80 in 8 guide covers whether it is still used in 2026.
Is the Citadel OA on HackerRank?
Yes, by consistent candidate reports: two to three algorithmic problems in 60 to 90 minutes at LeetCode medium-to-hard level, with some researcher tracks adding probability multiple choice. Full test-case passes matter a great deal given applicant volume.
How do I prepare for mental maths OAs?
Daily timed drills for two to six weeks: Quantt's 80 in 8 simulator for the penalty format, the mental math trainer for volume, and Zetamac if you want an extra untimed timer. Train with penalties for wrong answers so you learn when to skip, and track your net score rather than questions attempted.
Can I use a calculator or paper?
Calculators are almost never allowed on mental maths sections; that is the point of them. Rough paper is usually permitted but rules vary and many tests are webcam-proctored, so read each assessment's instructions before starting.
What happens if I fail a quant OA?
Usually a rejection for that cycle, with reapplication allowed after 6 to 12 months at most firms. Failing one firm's OA has no bearing on others, and many successful quants failed several OAs in their first season.
Do all quant firms use online assessments?
Most but not all. Jane Street has historically favoured going straight to phone interviews for trading roles, and some smaller funds screen by CV and referral. Everyone else on this page gates with some form of OA.
Practise the questions Quant Online Assessments: Firm-by-Firm OA Guide 2026 actually asks
Reading about the interview is one thing - sitting one is another. Open your free Quantt prep workspace for a real course lesson plus interview-style coding tests modelled on firms like Jane Street, Citadel, Hudson River and Optiver.
Free lesson + interview practice · No credit card required