Four weeks is enough to go from rusty to dangerous, provided every day has a job. Most failed quant interview prep does not fail for lack of hours. It fails for lack of structure: random brainteasers in random order, no system for revisiting misses, and no time pressure until the real interview supplies it. This roadmap fixes the structure problem and leaves you to supply the consistency.
The plan assumes you are targeting trading or quant roles where the interview spine is probability, expected value, mental math, and market-style games. It scales to your situation: there is a schedule for people prepping around a full-time job and one for students who can go deeper.
The three rules that make the plan work
Daily beats heroic. Forty focused minutes every day builds more reflex than a five-hour Saturday. Interview math is a motor skill as much as a knowledge base, and motor skills are built on frequency.
Misses resurface, always. A question you missed is the most valuable object you own, and it is worthless if you never see it again. More on this below, because it is the engine of the whole plan.
Pressure arrives on schedule. Untimed practice builds understanding. Timed practice builds the version of you that shows up in interviews. The plan introduces clocks in week 2 and never removes them.
Why your misses are the curriculum
Memory decays on a curve, and a single exposure to a hard idea does almost nothing durable. What flattens the curve is retrieval: being forced to reproduce the idea right when you are about to lose it, then again at a longer interval, then again.
A miss is also a diagnosis. It marks, with perfect precision, a spot where your intuition and the truth disagree. If that spot never resurfaces, you will re-miss it in the interview, except now with adrenaline involved. So every miss should come back at roughly one day, three days, and a week, until it solves cleanly twice in a row. The review queue schedules this automatically; your only job is to clear the queue daily and never mark a fuzzy answer as solved.
Your daily hour, or your daily three
| Block | With a full-time job (about 75 min) | Student or full-time prep (about 3 hours) |
|---|---|---|
| New questions | 25 min | 60 min, split in two blocks |
| Mental math sprint | 10 min | 10 min, twice |
| Daily set | 10 min | 10 min |
| Review of misses | 20 min | 30 min |
| Timed screen or mock | weekends | every other day |
| Out-loud re-solve of one old problem | 10 min | 20 min |
If you have a job, protect the morning sprint and the review block above all else; new questions can flex. If you are a student, resist the urge to triple the new-question block and skip everything else. Volume without review is how people study for four weeks and plateau in week 2.
Week 1: foundations
The goal this week is to make the probability core reflexive and to wake up your arithmetic. Nothing fancy. Champions of week 4 are built from boring week 1 reps.
The daily structure. Twenty-five minutes of new questions from /topics/probability/, working easy to medium. Then a ten-minute sprint at /games/mental-math/ to push raw arithmetic speed. Then the daily set, which takes about ten minutes and keeps your streak honest. Close with your review queue, which is small this week and will not stay that way.
What to cover. Counting and symmetry arguments, complements, conditional probability, independence, and the standard distributions of dice and cards. When a solution surprises you, do not just read it. Re-derive it from a blank page the same day.
Week 1 checkpoint. You should be solving easy-tier probability questions in a minute or two without notes, your sprint scores should be visibly trending up across the week, and you should be able to state Bayes' rule and explain a complement-counting argument from memory. If easy questions still take you five minutes on Sunday, extend this week rather than advancing on sand.
Week 2: expected value, market thinking, and the first clocks
Now the material starts looking like what desks actually ask, and time pressure enters.
Expected value. Work through /topics/expected-value/ with emphasis on linearity, which holds even without independence: . Add indicator variables, fair pricing of simple games, and the habit of asking what variance does to a bet that looks good on average. Pricing a game out loud, branch by branch, is the single most transferable interview skill.
Market-making concepts. Learn the vocabulary of quoting: bid, offer, width, edge, and updating when someone trades with you. You do not need market experience; you need the logic that width should track uncertainty and that flow is information. The tracks sequence this material so you are not guessing what comes next.
First timed screens. Take two short timed screens this week, one midweek and one on the weekend. The first score is a baseline, not a verdict. You are learning what the clock does to your accuracy, and that data shapes the next two weeks.
Week 2 checkpoint. You can price a multi-branch game out loud, start to finish, in about two minutes. You finished both screens without freezing, even where you ran out of time. Your review queue is being cleared daily, and week 1 misses are starting to resurface and fall.
Week 3: firm-specific focus
Generic prep has diminishing returns after two weeks. Now you aim it.
Pick targets. Choose two or three firms and read their pages at /firms/. Reported interview styles differ meaningfully: some firms lean on rapid-fire arithmetic screens, others on long conversational probability that keeps escalating, others on game play and group exercises. If Jane Street is on your list, the stage-by-stage breakdown and the firm page tell you what candidates report and what to emphasize.
Harder material. Move your new-question block to medium and hard problems in the question bank, filtered toward your target firms' style. Expect your hit rate to drop. That is the point: you are buying misses now so they cannot ambush you later.
Review discipline. This is the week the spaced-repetition engine pays out or stalls. Your queue at /review/ now contains two weeks of accumulated misses arriving on schedule. Clear it before touching new material, every day, even when it is the less glamorous half of your session.
Week 3 checkpoint. Hard questions produce structured attempts rather than blank stares: you can always name an approach within thirty seconds, even when you cannot finish. Your timed screen scores are stable or rising even as difficulty rises. Old misses now solve cleanly when they come back. If they do not, slow the intake of new material and let the queue catch up.
Week 4: simulation week
Stop learning new things. Spend the final week making the real interview feel like a rerun.
Mock loops. Run a full mock interview most days, out loud, under conditions that match what you expect: timed, spoken, no long silences. Treat each mock as a dress rehearsal, including the parts where you get stuck and have to narrate your way out.
Full-length screens. Two or three full timed screens spread across the week, taken at the same hour as your real interview if you know it. You are training your body clock as much as your math.
Calibration. Score your confidence, not just your answers. When you say you are 90% sure, check how often you are right. If your 90% intervals are catching the truth far less than nine times in ten, widen them and say so out loud in mocks. Interviewers reward candidates who know the size of their own error bars.
Sleep and logistics. Taper intensity across the week; the last 20% of cramming is worth less than the sleep it costs. Confirm the format with your recruiter, ask about paper and calculator policy, test your audio setup for virtual rounds, and plan to arrive or log in early. Boring logistics, handled early, buy you calm.
Week 4 checkpoint. Mock performance is consistent across days rather than swinging wildly. You can talk through a miss without spiraling. Your warm-up routine is fixed. You know exactly what you will do in the first ten seconds after hearing a hard question, because you have done it forty times.
The out-loud reasoning habit
Interviews grade the audio track. Two candidates with identical accuracy are miles apart if one narrates a clean model while the other goes silent for ninety seconds and announces a number. So from day one of week 1, solve out loud, even alone at a desk, even though it feels ridiculous.
Build a small set of spoken scaffolds and wear them in: naming the sample space before computing, stating a rough estimate before the exact attempt, flagging your confidence as a number, and announcing updates when new information arrives. Once a week, record a session and listen back. You are hunting for unstated assumptions, filler, and the silent stretches where your reasoning went private. The mock loop format exists precisely to make this habit social before the stakes are real.
The last 48 hours
Two days out. Your final heavy session: one full mock, review queue cleared, then stop. Cramming new material past this point trades calm for fragments you will not retain.
One day out. Light work only. A short confidence set of easy questions from the daily set, a single mental math sprint to stay loose, logistics double-checked, and a full night of sleep treated as non-negotiable training.
Day of. Fifteen minutes of warm-up an hour or two before: a few easy probability questions and one sprint, just enough to wake the machinery without burning glucose you will want later. Then stop, eat, and walk in ready to think out loud.
FAQ
Is four weeks really enough to prepare for quant interviews?
It depends on your starting point. With a STEM background and working knowledge of basic probability, four structured weeks is a credible runway, and it comfortably beats three months of unstructured puzzle browsing. If the week 1 checkpoint takes you two weeks to hit, stretch the plan to six weeks by doubling the foundation phase. Below two weeks of total runway, compress weeks 2 and 3 and protect simulation week; pressure-tested basics beat untested breadth.
What if I am still missing questions in week 4?
You will be, and that is fine. The signal to watch is trajectory and structure: misses that come with a clean attempted approach are normal even for strong candidates, while repeated blank stares at familiar question types mean you advanced too fast. Calibration matters more than perfection. A candidate who is right 70% of the time and knows it reads better than one who is right 80% of the time and claims 99.
Should I be preparing for coding interviews too?
Depends on the role. Trading roles lean heavily on the math spine this roadmap covers. Quant research and developer roles usually add programming rounds, so run your coding prep as a parallel track rather than carving it out of these sessions. The probability and expected value work transfers to every quant role; nothing here is wasted.
Start with day one, today
A roadmap you read is worth nothing; day one takes 35 minutes. Do the daily set, run one sprint at /games/mental-math/, and start the probability core at /topics/probability/. Then let the tracks carry the sequencing, the review queue carry your misses, and the mock loops carry the pressure. Four weeks from now the interview will not feel like a test. It will feel like a Tuesday.