How to Answer Brainteasers in Finance Interviews (Without Gimmicks)
A practical approach to finance interview brainteasers and estimation puzzles: structure, units, sanity checks, and when to push back.
IB · 4 min read
Some processes still throw brainteasers or Fermi estimates. The point is rarely the exact answer. It is whether you structure, keep units straight, stay calm, and narrate. Treat it like a mini case with worse data, not a trivia contest.
The short answer
"I clarify the question, state assumptions, build a simple equation, compute with round numbers, and sanity-check. I talk while I work. If the puzzle is ambiguous, I choose an interpretation out loud instead of freezing. I would rather be approximately right with clean units than precisely wrong with silence."
Default method
Clarify
What exactly are we estimating or solving? Restate it in one sentence.
Equation
Write drivers before numbers. Product of rates, shares, and prices beats jumping to a guess.
Guess openly
Round; label guesses as guesses. "About 330 million people" is better than fake precision.
Compute
Keep zeros under control. Track millions vs thousands out loud.
Check
Per person, per day, vs a known benchmark. If it implies every American owns three yachts, revise.
Need
Structure
Need
Units
Need
Sanity check
Finance-flavored examples
Common prompts:
- How many gas stations in the US?
- Market for office coffee in Manhattan?
- How would you value a bridge with tolls?
- How many piano tuners in Chicago? (classic Fermi)
These overlap market sizing for finance interviews. Use the same muscles: population or firms, penetration, frequency, price, then total. For a toll bridge, you are really doing a mini DCF or stabilized cash flow: traffic, toll, opex, CapEx, discount or cap rate. Say that. Bridging to valuation language is how you look finance-native instead of consulting-generic.
Worked sketch: gas stations
Clarify: operating stations in the US, today. Equation: drivers / stations per driver, or cars / cars per station. Say 330m people, 0.8 cars per person roughly, about 260m cars. Guess 1,000 to 1,500 cars per station depending on urban mix. Land near 170k to 260k stations. Sanity: does that feel high vs what you see on a highway corridor? Adjust. The interviewer cares that you built a path, not that you memorized a census number.
What interviewers dislike
- Silent staring for two minutes
- Fake precision ("exactly 147,382")
- Giving up after one wrong turn
- Ignoring impossible units ("dollars per gallon of people")
- Arguing with the premise instead of choosing an interpretation
Probability and logic teasers
For pure logic puzzles: restate rules, try a small case, look for invariants. If you have seen the exact puzzle, say so and still walk the logic. Do not smugly recite only the answer. Interviewers are grading process under pressure, not whether you read a puzzle book.
If it is a probability question, define the sample space, independence assumptions, and whether order matters. Wrong independence assumptions are a common fail even when arithmetic is fine.
When to push back
If a teaser has nothing to do with the job and the firm is known for them, play along with structure. If it feels abusive or unlimited, stay professional, structure what you can, and move to a bound: "I would put it between X and Y given these assumptions, and I would refine with one data pull if this were live work."
Do not lecture them that brainteasers are outdated. That is a values fight you will not win in a screen.
20-second opener
"I am going to define the target metric, outline an equation, estimate inputs with round numbers, and check whether the result is sane. Stop me if you want a different definition of the market."
Tie it back to real technicals
Brainteasers are seasoning, not the meal. Offers still turn on LBO walkthroughs, accounting, and valuation. After a teaser, be ready to pivot cleanly: "Happy to keep refining that estimate, or I can walk a three-statement or paper LBO if useful." That signal matters.
Practice sizing on public comps and simple markets the same week you drill accounts receivable +10 so your brain switches modes without panic.
Traps
- Starting with a number before an equation
- Mixing daily and annual rates
- Forgetting a price or frequency step in market sizing
- Over-apologizing for rough math (own the approximation)
- Spending the whole interview on a puzzle while technicals go untested: guide time back if they allow it
30-second verbal (meta)
"For brainteasers I clarify the ask, write a simple driver equation, plug round assumptions out loud, compute carefully, and sanity-check per unit or vs a known benchmark. Ambiguity gets an explicit interpretation, not a freeze. I treat it as structured estimation, not trivia."
Related
Pair with Market Sizing and core technical guides so one puzzle does not become your whole prep identity. On Project Catalyst, keep technical reps primary.
Practice on SheetRank
Apply what you learned with drills, guided Academy builds, and live deal underwriting.