The market making game in trading interviews

"Make me a market on the number of windows in this building." Once you are past the arithmetic screen, some version of this exercise appears in almost every trading interview. It looks like an estimation question. It is really a test of whether you understand what a two-sided quote commits you to, and whether you can keep a book straight in your head while someone trades against you.

Updated September 2026 · 8 minute read

How the game works

The interviewer names a quantity whose value is uncertain: a fact, the outcome of some dice, the sum of cards, anything. You quote a bid (the price you will buy at) and an ask or offer (the price you will sell at). The interviewer then trades with you: they can hit your bid (sell to you at your bid) or lift your offer (buy from you at your ask), usually for a stated size. Then you are asked to requote, and the process repeats. At the end, the true value is revealed and your profit or loss is settled against it.

Three things are being watched:

• Whether your first market is centred on a sensible estimate with a width that reflects how unsure you are.
• Whether you update correctly when traded against. If someone keeps lifting your offer, your price is too low.
• Whether you can state your position and P&L at any moment without being told. Losing track of your own book is the fastest way to fail this round.

The first quote

Start with an estimate and a confidence. If you think a building has about 400 windows and could believe anything from 250 to 600, a first market of 300 at 500 says exactly that. Two common mistakes:

Quoting too tight to look confident. A 390 at 410 market on something you barely know invites the interviewer to trade against you at size, and you will be wrong on one side. Width is not weakness. Width is honesty about your information, and you tighten it as you learn.

Quoting too wide to be safe. A 100 at 1000 market is not a market, and interviewers will say so. Nobody would trade with you, and a market maker who never trades earns nothing. A useful rule is that your first market should be one you would be mildly unhappy to be traded on at either side. If you would be delighted to be hit on your bid, your bid is too low.

Say the reasoning out loud. "About 40 floors, maybe 10 windows a side, four sides, so around 400, but I am not sure about the floor count, so 300 at 500." That sentence is worth more than the numbers.

Getting picked off

The interviewer is not a random trader. They usually know the answer, or at least know more than you. So when they lift your offer, the most likely reason is that your offer was cheap. This is adverse selection, and it is the central idea of market making: the people who choose to trade with you are disproportionately the ones who know your price is wrong.

The correct response is to move your market in the direction of the trade. If your 300 at 500 gets lifted, requote higher, perhaps 450 at 600. If it gets lifted again, higher still. Each fill is information. A candidate who gets lifted three times in a row and keeps quoting the same offer has told the interviewer everything they need to know.

How much to move depends on size. A one-lot trade is a hint. A ten-lot trade at your price is a shout. Move more for bigger fills.

Keeping the book

Every fill changes two numbers: your position (how many you are long or short) and your cash (what you have paid or received). If you sell 5 at 500, you are short 5 and have received 2,500. If you then buy 3 at 450, you are short 2 and your cash is 2,500 minus 1,350, which is 1,150. Your P&L once the true value V is revealed is cash plus position times V: 1,150 minus 2V. You should be able to say your position and average price after every trade, unprompted.

Two habits make this manageable under pressure. Keep a running position and running cash separately, never try to hold a running P&L. And after each fill, restate: "I am now short 2 at an average of 525." Saying it locks it in and shows the interviewer you have it.

Size

You will often be asked how many you will trade at your price, or the interviewer will name a size and ask if you are good for it. Bigger size means more P&L when you are right and more damage when you are wrong. Early in the game, when your market is wide and your information is thin, quote small. As your market tightens and you become more confident, size up. Declining to trade a huge size against someone who clearly knows the answer is a good decision, not a cowardly one, and interviewers reward it.

The probabilistic version

Instead of a fact, the quantity might be the sum of two dice, the number of heads in ten flips, or the value of a card. Now the fair value is computable: the expected value. Your market should straddle it, and its width should reflect the variance. For the sum of two dice, EV is 7 and the standard deviation is about 2.4, so a market of 6 at 8 is tight and defensible while 4 at 10 is generous. If the interviewer offers to trade many lots, remember variance matters more: a fair-value bet you take 50 times will be near its EV, a fair-value bet you take once is a coin flip with a large payoff attached. Being able to say the EV and the rough spread out loud is the whole exercise here.

What good looks like

1. A first market centred on a stated estimate, with width that matches stated uncertainty.
2. Moving your market in the direction of every fill, more for bigger size.
3. Position and cash stated correctly after every trade without prompting.
4. Tightening as you learn, sizing up as you tighten.
5. A clean settlement at the end: position, cash, final P&L, in that order.

None of that needs unusual intelligence. All of it needs repetition, because the failure mode under pressure is forgetting to do one of the five.

Practise it

Quickfire's market making simulator plays this game against you. You post a bid, an ask and a size; a bot trades against you when your price is wrong; you requote until you have bracketed the value, then you settle the book from memory and it checks your position, cash and P&L. There are rounds on real-world estimation, dice and cards, running a book, and EV and variance. Free, no account needed.

Make a marketMental maths guide