Market Microstructure Interview Questions
Market microstructure questions in quant interviews all orbit one idea: every order that arrives may come from someone who knows more than you, and the bid-ask spread is the price of that risk.
Master the Bayes update behind a two-sided quote and the rest of the topic — quoting, order books, spread estimation, multi-asset hedging — falls into place.
Last updated 2 July 2026 · sub-areas, difficulty mix and firm attributions on this page are compiled directly from the 49 market-microstructure problems in the QuantVault bank; firm tags are candidate-reported, not employer-verified, and no individual author is named in our source data.
- Core sub-areas
- Adverse selection & spread-setting · optimal quoting · order-book mechanics · spread & impact estimation · multi-asset market making
- Typical difficulty
- Skews hard — 1 easy, 19 medium and 29 hard across the 49-problem set
- Who leans on it
- Quant-trading loops above all — problems here carry candidate tags from Two Sigma, Citadel, HRT, Optiver, DRW and others
- Practice pool
- 49 problems in the bank · 5 free to open with the full worked solution
Where market microstructure shows up: trader onsites and market-making rounds — when an interviewer says “make me a market” and then trades against you, the follow-up questions are microstructure. For how it sits inside specific funnels, see the Two Sigma interview questions guide and the Citadel interview questions guide.
The surfaceThe market microstructure sub-areas quant interviews test
Five families cover essentially every microstructure question in the bank. Each row pairs a sub-area with the recurring question shape and a representative type — a flavor drawn from real problems, not a leaked wording — so you can see where the difficulty actually lives.
| Sub-area | Recurring shape | Representative type (flavor, not a real question) |
|---|---|---|
| Adverse selection & spread-setting | Condition on the trade itself: what does a buy tell you about value? | A Glosten-Milgrom setup where some fraction of flow is informed; derive the zero-profit bid and ask as conditional expectations via Bayes. |
| Optimal quoting against informed flow | Choose the quote that maximizes expected P&L against mixed flow | Pick a half-spread or a skew balancing fill frequency against losses to informed traders — usually a short one-variable optimization. |
| Order-book mechanics & price formation | Explain what the book does and why the price moves | A conceptual walk through limit vs. market orders, price-time priority, depth, and how one-sided flow turns into a price change. |
| Spread & impact estimation | Recover a hidden spread or impact from trade data | Show why bid-ask bounce makes successive trade-price changes negatively autocorrelated, then invert that covariance into a spread (the Roll estimator). |
| Multi-asset market making | Quote several correlated assets at once | Combine conditional updating with a correlation or hedging structure across instruments — the hardest tail of the set. |
What's confirmed vs. what varies: the sub-areas and the difficulty mix above come straight from the problem set, so they are stable. Which firm asks which flavor is candidate-reported through problem tags — treat the firm attributions as directional, and expect trading tracks to see far more of this topic than research tracks do.
The patternsSignature market microstructure question patterns
Three derivations generate most correct answers in this topic. Each worked box below is a 60–90 second micro-example in the interview's actual cadence — the reasoning template, applied to a generic setup rather than any firm's wording.
Adverse selection — the quote is a Bayes update
Takeaway: a market maker's ask is not “value plus margin” — it is the expected value conditional on someone lifting it. Quote conditional expectations and the spread appears by itself.
Shape. An asset is worth \(1\) or \(0\) with equal odds. A fraction \(\pi\) of traders are informed (they know the value); the rest buy or sell on a coin flip. Where do zero-profit quotes sit?
1. Condition on a buy. Informed traders buy only when the value is 1, so \(P(\text{buy}\mid V=1)=\pi+\tfrac{1-\pi}{2}\) while \(P(\text{buy}\mid V=0)=\tfrac{1-\pi}{2}\).
2. Apply Bayes. The zero-profit ask is \(P(V=1\mid \text{buy})=\tfrac{1+\pi}{2}\); by symmetry the bid is \(\tfrac{1-\pi}{2}\).
3. Read off the spread. Ask minus bid is exactly \(\pi\) — the spread equals the informed share of flow. The move: never quote value ± a guessed margin; condition on the trade and let Bayes set the width.
Break-even quoting — price the loss to informed flow
Takeaway: per fill you earn the half-spread from noise traders and pay the informed trader's edge — write that expected value down first, and the break-even quote falls out in one line.
Shape. You quote a half-spread \(h\) around fair value. Each arriving trader is informed with probability \(\pi\) and holds an edge \(e>h\); otherwise they are uninformed. What half-spread breaks even?
1. Write the per-fill EV. Noise pays you \(h\); informed flow costs you \(e-h\). So \(\mathbb{E}[\text{P\&L}]=(1-\pi)h-\pi(e-h)=h-\pi e\).
2. Set it to zero. Break-even is \(h^{*}=\pi e\): the half-spread must cover the informed share times their edge — toxicity priced as arithmetic.
3. Add the trade-off. Full interview versions layer a fill-probability curve that falls as \(h\) rises, turning the choice into a one-variable maximization. The move: EV per fill first, optimize second — never the reverse.
The Roll estimator — read the spread off bid-ask bounce
Takeaway: trades bounce between mid ± half the spread, so successive trade-price changes are negatively autocorrelated — and that covariance can be inverted into the spread itself.
Shape. The mid \(m_t\) follows a random walk; each trade prints at \(P_t=m_t+\tfrac{s}{2}q_t\) with \(q_t=\pm 1\) equally likely and independent. Estimate \(s\) from trade prices alone.
1. Difference the prices. \(\Delta P_t=\Delta m_t+\tfrac{s}{2}(q_t-q_{t-1})\) — the bounce term is the only part that repeats with opposite sign.
2. Take the autocovariance. With independent \(q\)'s and a random-walk mid, \(\mathrm{Cov}(\Delta P_t,\Delta P_{t-1})=-\tfrac{s^{2}}{4}\).
3. Invert. \(s=2\sqrt{-\mathrm{Cov}(\Delta P_t,\Delta P_{t-1})}\). If the sample covariance comes out positive, the estimator is undefined — say so out loud; interviewers listen for that failure-mode caveat. The move: model the bounce, difference it away from the walk, and the microstructure noise becomes the measurement.
Free practiceMarket microstructure practice questions by difficulty
The bank holds 49 market-microstructure problems; 5 are free to open with the full worked solution — deliberately one per sub-area above, so every row of the table has a problem you can actually try. The list is short on purpose: we would rather show five real problems than pad the page. One honesty note: the free pool under-represents the hard tail — 28 of the 29 hard problems sit in the paid tier.
Warm-up — easy (1 free)
- Order Book Dynamics and Price FormationOrder-book mechanicseasyfree
Core — medium (3 free)
- Glosten-Milgrom Zero-Profit Bid-Ask QuotesAdverse selectionmediumfree
- Optimal Bid-Ask Quotes Against an Informed TraderOptimal quotingmediumfree
- The Roll Spread EstimatorSpread estimationmediumfree
Stretch — hard (1 free)
- Cross-Asset Market Making with Correlated FundamentalsMulti-asset market makinghardfree
Done with all five? The remaining 44 problems — including nearly all of the hard tier — live in the full problem bank, and the market microstructure playlist sequences the whole set from warm-up to stretch.
The planHow to prepare for market microstructure questions
Five techniques to make reflexive, in the order they pay off. The first two alone cover the topic's most recycled interview derivation.
- Make the adverse-selection update your first instinct. Before any algebra, ask “what does this order tell me about value?” The ask is \(\mathbb{E}[V\mid \text{buy}]\), the bid is \(\mathbb{E}[V\mid \text{sell}]\) — everything else in the topic is decoration on that conditioning.
- Drill the zero-profit cadence end-to-end. Prior → likelihood of a buy under each value → Bayes → quotes → spread. Practice until the two-state version takes under two minutes on a whiteboard.
- Price toxicity as arithmetic. Per-fill EV = half-spread earned from noise minus informed share times edge. Be ready to bolt on a fill-rate curve and differentiate once — that single extension covers most optimal-quoting variants.
- Know the Roll estimator and its failure mode. Bounce implies negative autocovariance of price changes; invert to get the spread; positive sample covariance means the estimator is undefined. The caveat is part of the expected answer.
- Speak order-book fluently. Limit vs. market orders, price-time priority, depth, queue position, impact — the conceptual questions are the closest thing this topic has to free points. Then run the free practice set and take the interactive market microstructure playlist on a clock.
FAQMarket microstructure interview questions — frequently asked
How important is market microstructure for quant interviews?
Core for quant-trading loops, peripheral for most research loops: every problem in this set is tagged to trading-track interviews, and only a minority double as research questions. A clean market-making derivation is one of the strongest trader-signal answers you can give, because it shows you understand what the desk actually does.
What market microstructure topics should I focus on?
Adverse selection first: the Bayes update behind a zero-profit quote (the Glosten-Milgrom cadence) is the single most recycled derivation in the topic. Then the expected-value arithmetic of a quote against partly-informed flow, the Roll spread estimator, and enough order-book vocabulary to talk through priority, depth and impact.
How hard are market microstructure interview questions?
This set skews hard: of the 49 problems, 29 are hard, 19 medium and only 1 easy. The difficulty is almost never arithmetic — it is setting up the right conditioning, deciding what a trade tells you about value before you compute anything.
Are these real quant interview questions?
They are representative, not verbatim. The problems are curated from our bank to match the microstructure shapes candidates report, rewritten for clarity with worked solutions we author ourselves — we never claim any wording is a leaked question.