Quant Interview Market Making Roadmap
A market making interview roadmap for fair value, spread width, quote updates, inventory, and game review.
Quant trading candidates preparing for market making and trading game interviews.
Begin with fair value
Every market making game starts with a fair value estimate and an honest uncertainty around it, and the roadmap should train both in that order. Sometimes you can compute both exactly: asked to make a market on the sum of two dice the interviewer is about to roll, the fair value is 7 and the standard deviation is about 2.4, so a market of 5 at 9 is well calibrated and there is no virtue at all in going wider. More often the value is a judgment, in which case say the judgment out loud rather than only the number. An estimate with no reasoning attached cannot be updated intelligently later; it can only be abandoned.
Set spread from uncertainty
Width should be a consequence of your uncertainty, the adverse selection you face, and your inventory - not a personality trait. Two dice, where the distribution is known exactly, justify a tight market. An estimate you can only defend to within a couple of hundred does not, and a narrow market around it is simply a standing offer to whoever knows more than you do. Tighten only when the estimate genuinely improves or when the game rules cap how far you can be run over. Interviewers rarely punish a wide market with a clear reason, and they consistently punish a tight one without.
Update after trades
A trade is information, and the most common failure is reading your own fill as good news. If a counterparty is happy to trade against your quote, the first question is what they know that you do not. Move the midpoint toward the side that traded, and widen rather than narrow whenever you cannot explain why they traded at all. If the same side hits you three times running, your fair value is wrong rather than unlucky - and saying so out loud, before the interviewer does, is worth more than the trade cost you.
Concrete practice loop
Run a short game and log six fields for every quote: fair value, bid, ask, inventory, the trade, and the reason. Then review the largest decision error rather than the profit. Sessions work best when they isolate one skill - one session on nothing but initial width, another on reacting to one-sided flow, a third on quoting while already long or short ten units. That isolation is what makes the review specific enough to change the next quote.
Add communication
Market making interviews are live, so say why the market is wide, why it moved, and how inventory changes the next quote. "I am long eight already, so I am skewing my quote down five to attract sellers" is a complete answer; silently shifting the quote is not. A clear rationale routinely outscores a deceptively precise number, because the interviewer is hiring the reasoning rather than the guess.
Common mistakes
The failures here are failures of sequencing. Candidates practice quoting before they can estimate, so every market is a guess with a spread wrapped around it. They run game after game without isolating a single skill, so no session produces a lesson specific enough to act on. And they treat the last game as representative when a single trading game carries almost no signal - judge the width rule and the update rule across ten games, not the outcome of one.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.