The single most important concept in market-making interviews, and the one candidates most often lack.
The idea
You quote two-sided. You only trade when someone chooses to trade with you. That selection is not random - people trade against your bid when they think your bid is too high.
So conditional on being filled, your price was more likely wrong than it was unconditionally. Getting filled is evidence against your own quote.
The interview demonstration
An interviewer quotes you a market and lets you trade, then reveals the answer. Or the reverse: you quote, and they only ever hit your bid. If you notice that the one-sidedness is information and widen or lower your market, you have understood. If you keep quoting the same market and keep getting hit, you have not.
The signal to say out loud: "You've hit my bid three times running - I think I'm too high, so I'm lowering my market."
The winner's curse
The same mechanism in auction form. Bidders estimate a value with error; the winner is the one who overestimated most. So winning an auction is bad news about your estimate, and rational bidders shade their bids below their estimate by an amount that grows with the number of competitors.
Interviewers ask this directly: "You are bidding for an oil field along with nine other firms. Your geologist says it is worth 100 million. What do you bid?" Considerably less than 100, and being able to say why is the answer.
Where it shows up in practice
- Options market makers widen before events because informed flow concentrates there.
- Retail flow is valuable to internalise precisely because it is uninformed - the adverse selection is low.
- A desk whose fills are consistently followed by adverse price moves is being picked off and must re-price.
The response
Not to trade less overall, but to widen (charge more for the risk) or skew (move the quote away from the side you keep getting hit on). See inventory risk and skewing.
Practise in market making and the make a market game.