Brain teasers have a bad reputation, and deservedly so when used carelessly. But in quant interviews they are not random — they cluster into a handful of types, and once you recognize the type, most of the mystery disappears.
Why firms still use them
A brain teaser tests something a résumé cannot: how you handle a problem you have never seen, out loud, with someone watching. The answer matters less than the structure of your thinking. Interviewers are looking for whether you can state assumptions, simplify, and reason cleanly.
Type 1: Expectation by symmetry
The classic family: coin flipping until a pattern appears, waiting times, expected number of draws. The key insight is almost always that you do not need to enumerate everything.
Example shape: flip a fair coin until you see two heads in a row. What is the expected number of flips?
Strategy: define states (no progress, one head) and write linear equations for the expected remaining flips from each state. Symmetry and linearity turn intimidating problems into two lines of algebra.
Type 2: Betting and fair value
You are offered a game or a contract and asked what it is worth. These are disguised expectation problems.
Strategy: compute the expected payoff under the stated rules. Watch for the traps: optionality (you can stop early), adversarial opponents, and payoff structures that are actually options — where the value is driven by the maximum, not the average.
Type 3: Strategy games
Two-player games, splitting games, and auction-style puzzles.
Strategy: think backwards from the last decision. Check whether a player would ever be indifferent between two choices — indifference is where randomization enters. For splitting puzzles like the pirate game, write out each player's decision in turn, starting from the smallest effective group.
Type 4: Logic and enumeration
Weighing puzzles, hat problems, liar-and-truth-teller setups.
Strategy: find the information-theoretic bound first — each weighing has three outcomes, so n weighings distinguish at most 3^n cases. If the bound is tight, your solution is likely right. Otherwise, there is a cleverer encoding.
Type 5: Geometry and visualization
Shaded areas, angles in chains, and clock problems.
Strategy: add auxiliary lines, name unknown angles with variables and chase relationships, and look for the color or area that appears twice.
How to present your solution
- Restate the problem in your own words and confirm the rules. Interviewers reward this and it costs ten seconds.
- Try small cases. Compute the answer for tiny versions of the problem — patterns emerge fast and you get something to say immediately.
- Name your assumptions. If the problem is ambiguous, state your assumption and proceed. Stalling silently is the only true failure mode.
- Simplify ruthlessly. Rescale, relabel, throw away symmetries, or cut the problem to a special case you can solve.
- Generalize and check. Does your formula give sensible answers at the extremes?
A final note
If you have genuinely never seen a problem type and go quiet for three minutes, narrate what you are trying. "Let me try assuming the events are independent and see where that lands" is a perfectly good sentence. Silence is not.
Work a few problems from each type weekly — the brain-teaser and probability sets on the resources page map roughly one-to-one onto this taxonomy — and the interview versions will feel like old friends in new clothes.