Quant Interview Prep: Two Tracks, One Wrong Turn

You search "quant interview prep" and you get five articles that all say the same thing: study probability, read the Green Book, grind LeetCode, practice mental math. Most of it is accurate. None of it tells you that "quant interview" is actually two completely different tests wearing the same name tag.

Miss that distinction early and you spend six weeks drilling mental arithmetic for a Jane Street SWE role that has zero mental math. Or you grind dynamic programming for a trader role at Optiver and walk in underprepared on expected value. The fork is real, the advice online mostly ignores it, and it costs people real interview cycles.

The Fork Nobody Draws

There are two tracks into quant and HFT firms.

Track 1: Quant Trader / Researcher. You are buying and selling decisions, managing risk in real time, or running models that generate signals. Optiver, IMC, SIG, Flow Traders, Jane Street trading arm. The interview is probability-heavy with brain teasers, mental math under time pressure, market-making games, and expected value estimation. Coding exists but takes a back seat.

Track 2: Quant Developer / SWE. You build the infrastructure: execution systems, order management, market data pipelines, latency-critical networking. Jane Street SWE, HRT, Optiver SWE, Citadel Securities tech. The interview is predominantly a software engineering test, with the math component dialed way back.

Getting this wrong is the most common preparation mistake. Figure out which track you're on before you open a textbook. Before you read another blog post. Honestly, before you open this one, but too late now.

The Shared Foundation (Weeks 1-4)

Both tracks need this. Don't skip it.

Probability fundamentals. Not statistics from a data science course. Combinatorics, conditional probability, Bayes' theorem, expected value, and basic random variables. A trader interview might ask you nothing else for 45 minutes. A developer interview peppers in one or two probability questions as a sanity check. Work through the first two chapters of A Practical Guide to Quantitative Finance Interviews by Xinfeng Zhou (the Green Book) before doing anything else.

Core algorithms and data structures. Both tracks need you to code. Trader interviews may only ask you to implement a basic data structure or walk through a short problem. Developer interviews go much deeper. Either way, if you can't implement a hash map, binary search tree, or heap from scratch without looking anything up, you're not ready for either track. Grind 75 covers the necessary ground without burning three months.

Language fluency. Pick one language and know it cold. Python works for trader roles. C++ is the default for HFT dev roles. Jane Street lets you code in whatever you want, but you need to be fast and clean. Don't switch languages mid-prep. Switching languages mid-prep is the prep equivalent of rearranging deck chairs.

The Trader Track (8-12 Weeks Total)

Once the foundation is in, the trader track breaks into three stages.

Stage 1: Mental Math Is Not Optional (Weeks 1-3)

This is the filter most candidates blow through thinking they can get by. They cannot.

Flow Traders puts 80 questions on screen and gives you 8 minutes. Some firms do 30 in 5 minutes. Before you get to any brain teasers, you need to be fast on integer arithmetic, fractions, and percentage conversions. The benchmark that maps to passing these screens: score above 40 on ZetaMac with default settings. That number corresponds to the pace at which mental math stops being a bottleneck during live problem-solving.

Fifteen minutes a day, every day, for six weeks. That's the minimum dose. Use ZetaMac. Don't use a calculator. The first week, you will feel genuinely stupid. That's correct. Keep going.

Stage 2: Brain Teasers and Probability (Weeks 3-8)

The Green Book is your primary resource. For trader roles, focus on chapters 1 (probability theory), 2 (brain teasers), 4 (stochastic processes), and 5 (finance fundamentals). Heard on the Street by Timothy Crack is a useful supplement with different problem sets and clearer explanations on some topics.

The key thing interviewers are scoring is your verbal reasoning, not just the final answer. When an Optiver interviewer asks for the expected value when you roll two dice and pay the sum in dollars, they want to hear you derive it out loud. A silent pause followed by "7" is worse than a slightly wrong answer with clear step-by-step reasoning. "7" delivered in silence reads as a lucky guess. Work through it.

Classic problems to master before showing up: the pirate game, the hat color puzzles, expected number of coin flips to get heads-heads, the optimal stopping secretary problem, and problems involving Bayesian updating. Aim for about 80-100 brain teasers across the two books. The top quant and HFT coding interview problems page has a list organized by firm if you want to drill by target.

Stage 3: Market-Making Games and Finance Intuition (Weeks 8-12)

The last stage actually simulates the job. Market-making games like the ones on Tradermath put you in the position of quoting a two-sided price on a random variable and then defending your market as the interviewer trades against you. "I roll a die, pay you the sum of the faces. Make me a market." The answer is not just "3.5 bid, 4.5 offer." The interviewer will hit your bid repeatedly. You need to widen and move.

Spend a few weeks here after the math is solid. Running this exercise under pressure is what separates someone who studied trading from someone who has actually internalized it.

You also need a basic grip on market microstructure: what bid-ask spreads represent, why market makers need edge over expected value, how inventory risk works. You don't need options pricing theory. You need intuition.

Readiness signal for the trader track: you can work through a medium-difficulty probability problem out loud without pausing, your ZetaMac score is above 40, and you have run at least 20 market-making practice sessions.

The Quant Developer Interview (8-16 Weeks Total)

The developer track has its own split. Jane Street SWE is fundamentally different from HFT SWE at firms like HRT or Optiver.

Jane Street SWE: Iterative Problem Building

Jane Street interviews are collaborative coding sessions, not LeetCode runs. The typical structure is a single problem in three parts, each building on the previous. The canonical example from their own blog walks through implementing memoization, then adding FIFO eviction to fix a memory leak, then upgrading to LRU eviction. All in one 60-minute session.

Every candidate should complete Part 1 bug-free and quickly. Most candidates who advance finish Part 2. Completing Part 3 is a strong signal. Not a hard requirement for passing, but it tells the interviewer you think architecturally, not just mechanically.

Practicing classic problems from scratch under time pressure matters more than memorizing solutions. They explicitly ask candidates not to share questions because they're testing problem-solving, not recall. Implement connect four, a text editor with undo, a basic JSON parser, an event queue. Problems with real structure that build on each other.

The mental math drills and brain teasers that dominate the trader track are irrelevant here. If you spent two weeks on the Green Book for a Jane Street SWE offer, you wasted two weeks.

Four-panel cartoon: interviewer asks "Tell me about yourself..." and the candidate beams "I develop code and test it!" The interviewer asks "What else?" and the candidate sweats while the interviewer stares blankly: "That's not your whole personality, right?"

Every candidate who spent 10 weeks on Optiver mental math and then applied to Jane Street SWE.

HFT SWE: Systems and Latency Awareness

At HRT, Optiver (SWE track), and similar firms, the interview adds a layer of systems knowledge that standard FAANG prep does not cover.

The core test is whether you think about hardware costs when you write code. Cache line size (64 bytes), false sharing between threads, the difference between L1 access (2ns) and DRAM access (80-100ns), why sequential memory access runs 5-10x faster than pointer chasing regardless of algorithmic complexity. These come up naturally in system design problems: "Design a market data feed that processes a million events per second." A candidate who thinks only about algorithmic complexity gets a very different answer than one who also thinks about memory layout and branch prediction.

For this track, add a 4-6 week C++ systems block after your DSA foundation. C++ Concurrency in Action by Anthony Williams is the canonical reference for threading and lock-free programming. Study the memory hierarchy, understand RAII and move semantics, and know what false sharing looks like in practice. Optiver specifically asks about concurrency, ring buffers, and lock-free stacks. The C++ for coding interviews foundations apply, but HFT takes it a level deeper into memory ordering and hardware awareness.

Optiver also runs a cognitive battery called Zap-N: nine browser-based games testing reflexes, pattern recognition, and decision-making under time pressure. You can't practice-prep it in the traditional sense, but a week of similar puzzle games beforehand helps with pacing.

Readiness signal for the HFT SWE track: you can implement an LRU cache cleanly in C++ in under 30 minutes, explain why a spinlock is better than a mutex in specific scenarios, and design a low-latency market data ingestion system with concrete trade-offs on memory layout and thread pinning.

Your Resume Gets Filtered Before a Human Sees It

Both tracks have a resume problem that prep guides skip.

Quant firms weight competitive programming backgrounds heavily. An IOI or ICPC finalist gets fast-tracked past the online assessment at some firms. An ICPC semifinal finish, a Codeforces rating above 1800, or a top math olympiad finish all read as strong signals. Not a requirement. But if you have it, put it at the top. The transfer from competitive programming to coding interviews is real, but what quant firms test beyond competitive programming is a different axis entirely.

If you don't have that background, a relevant project helps. An order book simulator. A market data feed parser. A Monte Carlo pricer. Something that proves domain interest and systems thinking beyond coursework. "I built a todo app" is not that thing.

The Timeline That Actually Holds

Starting from solid math and decent coding: the trader track takes about 10 weeks at two hours per day. Jane Street SWE takes about 8. HFT SWE takes 12-16 weeks to get the systems depth in.

If you're starting from a weaker math background or haven't coded seriously in a few years, add six to eight weeks to any of those.

Most applications open in July and August. Apply the moment listings go live. Not when you feel ready. The moment they go live.

If you want to stress-test your problem-solving out loud before the real thing, SpaceComplexity runs voice-based mock interviews with rubric-based feedback. It is built for DSA practice, but the core skill transfers: reasoning under pressure, narrated so someone else can actually follow it.

Recap

Two tracks. Identify yours before you start.
Shared foundation: probability basics, core DSA, language fluency.
Trader track: mental math first (ZetaMac above 40), brain teasers (Green Book chapters 1, 2, 4, 5), then market-making games.
Jane Street SWE: iterative problem-building. Not LeetCode, not mental math.
HFT SWE: memory hierarchy, C++ concurrency, cache-conscious design.
Trader track takes 10-12 weeks from a solid baseline. HFT SWE takes 12-16.
Competitive programming credentials help. Put them on your resume.
Apply when listings open.