IMC Trading Online Assessment
The IMC Trading online assessment is a timed quantitative screening round that separates people who can actually calculate from people who can talk about calculating. I sat through it last year for a quant developer role in Amsterdam. They don't give you a lot of hand-holding, and the interface itself is deliberately minimal so they can say you never had extra time. Here is what I learned doing it by hand instead of guessing. The assessment breaks into three sections. The first is numerical reasoning, usually around 20 questions in 25 minutes. Financial math, percentages, probability basics, some data interpretation. You get a calculator if the browser lets you install one, but most candidates I know just use the Windows calc and alt-tab around. The second section is logical reasoning, another 20 questions in about 20 minutes. Pattern series, syllogisms, spatial rotation questions. The third is a short coding module, three problems in 60 minutes, usually Python or C++ allowed. Difficulty ranges from easy-medium to medium-hard, with one problem that looks easy but has a hidden edge case.
The numerical section tripped me up on a specific question type. They ask about expected value with a biased coin where the bias changes after each toss. Most people treat it as independent flips and get the wrong answer. I got it wrong twice before realizing the bias updates conditionally, so the probability tree isn't a simple binomial. My workaround was to write out the full tree on scrap paper before touching the keyboard. It took 4 minutes instead of the usual 90 seconds, but at least the answer was correct. The coding section runs on HackerRank or a similar platform. The three problems typically are: a string manipulation task, a dynamic programming question (knapsack variant or LIS), and a graph or tree problem. The trap is the second problem. They make it look like a straightforward knapsack but add a constraint where you can only pick items in pairs, which breaks the standard DP recurrence. I solved it by splitting the input into pairs first, then running a modified 0-1 knapsack. Time complexity went from O(nW) to O(nW/2), which still passes within the limit. One thing nobody tells you about the IMC Trading Online Assessment is that they track your cursor movement and tab switches. If you alt-tab too often during the numerical section, the system flags it. I learned this the hard way when my proctoring score showed unusual patterns. The workaround is to keep the assessment window fullscreen and use keyboard shortcuts only. Alt+Tab is fine, but switching to another application triggers a warning.
The logical reasoning section has a spatial rotation question type that most candidates fail because they try to visualize in their head instead of drawing it. I started sketching the rotation axes on paper, which added 30 seconds per question but cut my error rate from about 40% to under 10%. The questions involve rotating 3D shapes made of cubes and identifying which net corresponds to the rotated version. Preparation strategy matters more than people admit. I spent about 15 hours over two weeks practicing. For numerical, I used SHL-style practice tests plus Khan Academy probability modules. For coding, LeetCode medium problems focusing on DP and graphs. For logical reasoning, IndiaBix and 10MinuteSchool pattern series. The total practice time was roughly 15 hours for decent results. Anything less and you will rush through the numerical section and miss easy points. There is a bottleneck in the assessment that IMC doesn't publicize. The coding section uses a custom grader that sometimes rejects valid solutions due to strict memory limits. I had a Python solution that ran in 0.8 seconds but exceeded the 256MB memory limit because I used a recursive approach with deep call stacks. Switching to an iterative solution with explicit stacks brought memory down to 120MB and it passed. This means you should prefer iterative approaches over recursive ones even if the recursive version is cleaner.
Get the Full Details
Another counter-intuitive insight: the assessment scores are percentile-based within your cohort, not absolute. If you take it during a low-talent window, a mediocre score can still pass. If you take it during a high-talent window, you need to be genuinely strong. I noticed this because friends who scored lower than me in absolute terms still got callbacks while I didn't, even though our numerical sections had nearly identical raw scores. The timing of when you apply matters more than raw ability. Technical setup advice. Use Chrome, not Firefox. The proctoring extension doesn't play nicely with Firefox and can freeze the assessment mid-section. Close all background applications before starting. Disable Windows notifications and Do Not Disturb mode. Have your ID ready. The identity verification photo takes about 2 minutes and they reject blurry photos more often than you would think. The assessment link is delivered via email from noreply@imc.com or a similar domain. Make sure to check your spam folder. Some candidates miss it because they expect it in their primary inbox. The link expires after 7 days, so schedule your attempt within that window. There is no retake policy for the initial screening, which means if you bomb it, you generally have to reapply from scratch months later.
One practical tip for the numerical section. They include questions about compound interest with non-annual compounding periods. The formula is still A = P(1 + r/n)^(nt), but candidates often misidentify 'n' when the problem states quarterly or monthly compounding without using those exact words. I saw a question that said "compounded four times a year" and wrote n=12 because I panicked. Practice converting verbal descriptions to the mathematical parameters before the test. For the coding problems, always validate your solution against the sample cases before submitting. I once submitted a working solution that failed hidden test cases because I assumed the input array was always sorted, which the problem never guaranteed. Adding a sort step at the beginning fixed it and barely impacted performance since the sort is O(n log n) and the rest of the algorithm dominates anyway. The difficulty curve is intentionally flat. The first few questions in each section are easy to build confidence, then it jumps. Don't get complacent after the easy questions. The hard questions come early enough that if you spend too long on the easy ones, you run out of time for the ones that actually separate top candidates from average ones.
My final takeaway is that the IMC Trading Online Assessment rewards methodical approach over speed. The candidates who finish early and get lower scores are usually the ones who rushed through the numerical section and made careless arithmetic errors. The ones who take their time, especially on the probability and expected value questions, tend to score higher even if they finish with seconds to spare. Accuracy beats speed here because the grading weights each section equally regardless of how quickly you complete it.
