# What a market-making interview actually looks like

Source: https://www.techinterview.org/post/3233476306/market-making-interview-trading-games/
Updated: 2026-07-13 · techinterview.org

"I'm thinking of a whole number between 1 and 100. Make me a market, width no more than 10." You have about three seconds before the pause gets awkward. Say "45 at 55" and the interviewer will either sell to you at 45 or buy from you at 55, and then they start handing you clues. That exchange, run for twenty minutes with variations, is the spine of a market-making interview.

Optiver, IMC, SIG, Flow Traders, and Jane Street all run some version of this for their trading seats. The costume changes between firms, the thing being measured does not: can you put a price on something uncertain and stay solvent when the market tells you that you were wrong.

## What it means to make a market

Quoting a two-way price means naming a bid, the price you will buy at, and an offer, the price you will sell at, at the same moment. The gap between them is your spread. If you believe the fair value of some quantity is 50 and you quote "45 at 55," you are willing to buy at 45 or sell at 55, and you have given yourself a five-point cushion on either side of your guess.

The catch is that you do not pick which side trades. The interviewer does, and they take the side that is good for them. Quote too tight around a number you are unsure of and you get picked off every time your estimate is off. Quote too wide and you look scared and never capture any edge. The entire game lives in that tension.

## The mock market, round by round

A typical sequence goes like this. The interviewer names a quantity you cannot know exactly. Sometimes it is playful, the number of piano tuners in Chicago, the count of countries in Africa, the sum of three dice about to be rolled. Sometimes it is dressed as a market, the price of a stock right after an earnings miss. You quote. They trade. Then they reveal something, the number is even, one die came up six, the CEO just resigned, and you requote on the spot. P&L is tracked across the whole exchange, and at the end they may ask what your position is and what it is worth.

Say the quantity is the sum of three standard dice. The mean is 10.5, so a reasonable opening market is "9 at 12." The interviewer lifts your offer, buying from you at 12, so you are now short one contract. Then they roll one die and it shows 6. The two hidden dice are still worth 3.5 each on average, so the expected total is now about 13, up from 10.5. Your market has to climb with it, say "12 at 14." Leave the stale "9 at 12" quote sitting there and the interviewer will gladly buy again at 12 something now worth 13, and you have handed over free money.

## Where the dice and card games fit in

Not every station is a live market. SIG leans hard on card and dice games built around expected value: you get a payout rule and have to say whether you would take the bet, at what price, and how much you would put behind it. One classic runs like this. A card comes off a standard deck and you win its value in dollars, with face cards counting as ten and the ace as one. What is a fair price to play? The average card is worth about 6.5, so paying 6.5 is break-even and paying 5 is a bet you should take again and again. Then they twist the rules, now you can pay to discard the first card and draw a fresh one, or you draw two and keep the higher, and each change shifts the EV while they watch you recompute without reaching for a calculator.

Optiver's screen opens with raw arithmetic before anyone talks strategy: eighty questions in eight minutes, no calculator, negatives and decimals mixed in, and a passing bar most people find rough on the first try. Nobody believes traders do long division all day. The test is whether 7.5 times 40 stalls you for two seconds, because if it does, you cannot hold a market and track a position while someone is actively trading against you.

## What the game is really scoring

| Interview round | Firms known for it | What it is really scoring |
| --- | --- | --- |
| Timed mental math (e.g. 80 questions in 8 minutes) | Optiver, Flow Traders | Arithmetic speed and error rate under a clock |
| Make a market on an unknown quantity | Jane Street, IMC, Optiver | Pricing uncertainty, choosing a spread, updating on new information |
| Expected-value card, dice, and coin bets | SIG, Jane Street | EV calculation, edge recognition, bet sizing |
| Live trading game with order flow | IMC, Flow Traders | Inventory management, adverse selection, staying calm under P&L swings |
| Probability brainteasers | All of the above | Structured reasoning with partial and conditional information |

Underneath the formats, the same handful of traits get graded. Mental math is table stakes. The one that separates people who have thought about this from people who have not is adverse selection. Your counterparty only trades when they think your price is wrong in their favor, so the flow you get is skewed toward the moments your estimate is off. A market maker who quotes a perfect midpoint with a razor-thin spread still bleeds money, because they buy right before the number falls and sell right before it climbs. Widening the spread is how you charge for that risk. As the interviewer feeds you real information and your uncertainty shrinks, you tighten, less unknown means you can afford a narrower quote.

Inventory is the other half of it. If you have bought three contracts you are long and exposed to the price falling, so a good trader skews the next market down, showing a lower offer to sell the position back toward flat and a lower bid to stop accumulating. Quote the same symmetric market every round no matter what you are holding and you lose marks, because you are ignoring the risk already sitting on your book.

And you have to say all of it out loud. These rounds are run as conversations by design. The interviewer wants the reasoning: why the width is where it is and what the last clue just changed. Silence reads as freezing or guessing. Someone who says "that six pushes my fair up to about 13, so I'll show 12 at 14, and since I'm short I want to buy some back, so I'll lean the quote up" is showing the exact skill being hired for.

## The mistakes that sink strong candidates

The most common failure is not bad arithmetic, it is going quiet. People try to compute the perfect number and freeze, when a fast approximate market beats a slow exact one every time. Close behind is stubbornness, clinging to an opening quote after three pieces of news should have moved it. Then there is over-trading, taking every bet because the EV looks positive on paper, without noticing the position has grown big enough that one bad roll erases a day of small edges. Sizing is part of the answer, not a footnote to it.

A subtler error is treating the interviewer as an opponent to outsmart rather than a market to read. They are telling you something through what they choose to trade. If they keep lifting your offers no matter how high you move them, that is a message: fair value is above where you think it is, and you should stop selling.

## Preparing without a trading background

None of this needs finance coursework. Get your mental math genuinely fast, especially multiplication and taking percentages of round numbers, until it runs in the background. Drill expected value on cards and dice until pricing a bet is reflexive. Play the free market-making simulators that copy the Optiver and Flow Traders formats, and practice narrating while you do it, because doing the math silently and doing it while you explain are different skills that fail in different ways. Poker and sports betting are decent training too, since both force you to price uncertainty and size bets with your own money on the table.

The firms running these games are not hunting for the highest IQ in the building. They want someone who can be wrong cheaply, adjust fast, and stay level while money moves against them. The number between 1 and 100 was never the point. How you priced it, and what you did once they traded, is the whole test.