Your IMC Trading Prep Plan

You've seen how IMC's software loop is built and judged. Now turn that into a plan. Two facts set the order. IMC's own interviewer has called the HackerRank round "purely theoretical, just a time-saving filter." The real signal lives in a working session that simulates the actual job, graded on how you work rather than the polished end product. And this is the software-engineer stream. The BrainsFirst games, mental-arithmetic tests and market-making Superday you may have read about belong to the trader pipeline. So the plan runs in the loop's order: the assessment first because it's a gate, then the working session and CS fundamentals where the loop is won, then the video and values.

The IMC Trading angle

Before you drill anything, settle two facts. First, your format, which varies by role, region and intake. Most candidates sit a HackerRank online assessment of roughly two hours with two medium-to-hard coding problems; some intakes add a handful of multiple-choice questions or an object-oriented round. A few regions and roles hand you a four-hour take-home instead: code a small game, judged on requirements, testing and documentation. Read the interview process and how hiring decisions are made first. Second, your level. A new-grad loop centers the working session and the CS-fundamentals round, while an experienced loop can add a classical low-latency system-design round, so prep the blocks your level faces.

• The filter: a HackerRank assessment with signature shapes. roughly two hours, two medium-to-hard problems, in C++, Java or Python. The shapes recur across intakes. Expect a grid-pathfinding Does-the-Path-Exist search solved with BFS or DFS and memoization, a per-second Premium question queue simulation, and a data-structure design problem such as an O(1) stack that also returns its sum. Clearing it buys the loop. It isn't the real signal
• The differentiator: the 2.5-to-3-hour working session. two senior engineers hand you a problem on an approximation of a real trading system. You read it alone for about thirty minutes, then solve it together with a hard focus on performance and alternatives. They judge problem-solving, how fast you pick things up and how you reason about trade-offs. A finished product matters less. Almost nobody drills for this shape, so it separates people
• The frame: CS fundamentals and a concrete why-IMC. a 1:1 theory round probes threads, memory, hashing, sorting, signed versus unsigned, and what happens when you type a URL, in a low-latency register. The one-way video and the recruiter call test fit against the six values. They expect you to know what IMC does and why you want it, and a generic line about wanting to work in finance falls flat

The plan, phase by phase

Each phase has one focus and a checkpoint that tells you when you've earned the next. Phases are shares of your prep time, so the plan holds whether you have a month or one hard week. The assessment comes first because clearing it earns the rest of the loop, but it's a filter, so it gets less time than its position suggests. The working session and the CS-fundamentals round take the largest share, because that's where IMC decides. Low-latency system design scales with your level, and the video and values run lighter, though they're a real part of the read. Don't step past a checkpoint until you can clear it without notes.

Picture a three-week runway. The first days settle your stream, level and assessment format, so you aren't drilling trader games or a round your level never faces. Week one goes to the HackerRank assessment: the grid-pathfinding search with BFS or DFS and memoization, the per-second Premium question simulation, and a data-structure design problem, all against a clock. Week two is the heaviest. Split it between the working session (frame an unfamiliar problem alone, then solve it out loud with a partner) and the CS-fundamentals round on threads, memory, hashing and sorting. Save the last stretch for a level-scaled pass at low-latency system design and for the one-way video, with value-mapped stories and a why-IMC you can deliver cold. On a shorter runway keep the order and shrink each block, because the assessment earns the loop and the working session wins it.

The IMC Trading angle

Take IMC at its word about what it measures. Its own interviewer calls the HackerRank round a time-saving filter and puts the signal in the working session. So the failure mode is treating the assessment as the whole game and walking into the working session cold. That round watches how you approach an unfamiliar problem. Do you use the read-alone time to structure it, think out loud with a colleague, weigh alternatives, and make performance part of your reasoning? Someone who has ground a hundred timed mediums but never spent thirty quiet minutes framing a hard problem will lose where this loop looks hardest. Practice the working session as its own skill, out loud, with a partner, on problems big enough to need structuring first.

Why the working session decides it

The weighting tracks where this loop separates people. You have to clear the assessment, but IMC has said plainly it's a filter. Once you can code the signature shapes cleanly, puzzle count stops distinguishing you. The working session and the CS-fundamentals round do. The working session runs two and a half to three hours, with two senior engineers watching how you reason through an unfamiliar problem on an approximation of a real system. The fundamentals round pushes on threads, memory, hashing and sorting in a low-latency register that a stack of medium graph problems never prepares you for. So the plan puts its largest blocks there instead of a fourth set of assessment puzzles, and scales system design to your level.

In the room

• Confirm the stream and format before you drill. This is the software path. Most candidates get a roughly two-hour HackerRank assessment with two problems; a few regions substitute a four-hour take-home. Prep the round your level and req faces.
• Clear the assessment on its signature shapes. Expect a grid-pathfinding Does-the-Path-Exist search (BFS or DFS with memoization), a per-second Premium question simulation, and a data-structure design such as an O(1) stack with sum. On a Python req, watch for time-limit errors on heavy inputs.
• Give the heaviest block to the working session. Two senior engineers bring a problem on an approximation of a real trading system. You read alone for thirty minutes, then spend an hour or more solving it together, focused on performance and alternatives. They grade how you work, so think out loud and name trade-offs.
• Own the CS fundamentals in a low-latency register. Be fluent on threads and concurrency, stack versus heap, how a hash table works, quicksort's trade-offs, signed versus unsigned, and the URL-to-render walk. Concrete, performance-aware answers beat framework trivia.
• Bring value-mapped stories and a concrete why-IMC. The one-way video mixes behavioral prompts with technical agree-or-disagree statements, so rehearse to camera. Map stories to the six values in your own words, and tie why-IMC to the market-making, low-latency work.

The IMC Trading angle

The trap is preparing for the wrong IMC. Spend a month on the nastiest dynamic-programming problems and you've aimed at a round the firm itself calls a filter, while the working session and fundamentals go unrehearsed. Spend hours on mental-arithmetic drills and market-making games and you've prepared for the trader pipeline. Good preparation is specific. Code the assessment's signature shapes cleanly and rehearse the working session out loud until framing a problem alone, then solving it together, feels natural. Answer the fundamentals in a performance-aware register, do a level-scaled pass at low-latency design, and know honestly why you want IMC. Put your marginal hour into the working-session habit you haven't built or the concurrency answer you fumbled, and skip the fifth hard puzzle.