Almost every top trading firm will put a puzzle in front of you, and how you approach it usually matters more than whether you land the exact answer. The questions are deliberately unfamiliar, because an unfamiliar problem is the only way to watch someone actually think.
This guide covers the three families of question that come up, how to work them under time pressure, and how to train for them.
Why firms ask brainteasers and probability puzzles
Trading is quick, logical decision-making under uncertainty, and these questions are the cheapest way to test for it. They are tough and unfamiliar on purpose, which is what exposes your raw approach: whether you are methodical, whether you stay structured when you are stuck, and whether being challenged rattles you.
Probability and expected value dominate for an obvious reason. Traders reason about odds and expectations continuously, so an interviewer wants to see you do it in real time. Firms often drill deeper and deeper into one problem, or add complications mid-answer, specifically to see whether your reasoning stays consistent under strain.
None of it is a test of finance knowledge. It is a test of how you think when you do not already know the answer.
The question types
| Type | What it looks like | Typical examples | Practise |
|---|---|---|---|
| Probability and expected value | Computing the probability of an event or the expected value of a random outcome | Expected number of flips until the first head; the chance of drawing 2 red cards from a deck; the two-thirds of the average game | Probability questions and the cards and coins collection |
| Brainteasers and logic | Classic riddles, pattern puzzles and lateral problems with no obvious formula | The mislabeled jars; the light bulb switching problem; why manhole covers are round; find the next term in a sequence | Reasoning puzzles and the top 50 |
| Statistics and estimation | Basic statistics, and Fermi problems where you estimate with almost no data | The birthday paradox; how many piano tuners are in New York; how many $1 bills would fill this room | The Fermi trainer and the Fermi estimation game |
Why each matters. Probability questions check that you understand independence, conditional probability, combinatorics and expected value, which are the working tools of pricing risk. Logic puzzles check whether you can decompose something you have never seen, which is what spotting a market inefficiency actually feels like. Estimation checks whether you can get to a defensible order of magnitude without a calculator, which traders do constantly.
Two categories sit outside this guide but do come up: game theory and betting games, and pure mental arithmetic drills. For those, see our make me a market guide and the mental arithmetic guide.
Solving under time pressure
Think out loud
Do not solve in silence. Restate the question in your own words, say how you plan to approach it, then narrate as you go. It keeps you organised, it lets the interviewer follow you, and it is the only way they can hand you a hint when you drift. It turns the session into a discussion rather than an interrogation.
Stay in the problem
If you hit a wall, say which part is hard rather than going quiet. Try a smaller case: shrink a probability problem to two or three trials, find the pattern, then scale back up. If you make a mistake, correct it and move on without apologising at length. Composure under difficulty is part of what is being marked.
Know when to approximate
If a calculation is ugly and precision is not the point, say you are estimating and do it: 17 × 93 is 17 × 100 minus 17 × 7, so about 1,581. Nobody penalises a stated approximation. In timed written tests there is usually no penalty for a wrong answer either, so never leave a question blank.
Take the hint
Interviewers drop hints and ask leading questions deliberately, to see what you do with them. Acknowledge the hint, adjust, and carry on. Ignoring guidance reads badly, because a trading floor runs on learning fast from people more senior than you.
Impose structure
Write down what you know and what is being asked. Name the type of problem: combinatorics, geometry, or a trick question. Rephrase it, or draw it. If it has several moving parts, solve a smaller version first. Structured thinking under a clock is precisely the signal they are looking for.
How to train
Work real questions, not textbook ones
Exposure to problems that firms have actually asked is what builds the pattern library. Our interview question database collects them from top trading firms, filterable by firm and category, and grows as candidates contribute new ones. Themes repeat more than you would expect: coin-flip expectations, dice odds, a handful of classic riddles. Time yourself from the start, since solving slowly is a different skill from solving.
Start with the essentials for coverage, then the top 50 for the ones that recur most.
Practise under real conditions
Most processes open with a timed online assessment, so train against a clock rather than at leisure. Set up 20 assorted questions in 20 minutes. Then practise the live half separately: have someone ask you puzzles and talk through your solutions out loud, or record yourself doing it. It feels awkward and it is the single highest-return thing you can rehearse.
Review every problem you get wrong
After each question, work out whether you could have got there faster or more simply. Note the heuristics as you find them: the expected number of flips until heads is 2, Monty Hall is a conditional probability problem once a door opens. That accumulated playbook is what makes the next unfamiliar problem tractable.
Attack your weak spots
You will find categories that consistently catch you. If card problems are the issue, review combinations and permutations and work through card draws until they are routine. If estimation is weak, do Fermi problems until decomposition is a reflex.
Use the rest of the toolkit
If your process includes a timed assessment, drill the format itself: the Math Trainer for arithmetic and the Sequence Trainer for pattern questions. For the interactive rounds, the market games drill quoting and expected value against a live position. When you know where you are applying, our firm pages collect the process and the practice for each firm, so you can work only the questions reported there.
Final thoughts
Drilling puzzles trains two things at once: thinking quantitatively, and staying composed in front of a problem you do not recognise. Firms are selecting for both, and the second is harder to fake.
Preparation is not optional here. Many successful candidates work at this daily for months, and the payoff is not only a better answer but a calmer one. Once you have solved dozens of hard problems, a new one in an interview stops being a threat and becomes familiar territory.
For how these questions fit into the wider process, read our full guide to quant trading interviews.