The Economics Of Market Making

A market maker quotes both a bid and an ask, hoping to buy low and sell high on volume that has nothing to do with any forecast of its own. Post a bid at 99.99 and an ask at 100.01, get filled on both, and you've captured $0.02 per share for doing nothing but standing in the middle. Repeat that ten thousand times a day and it looks like free money. It isn't, because two costs eat directly into that spread: inventory risk (you don't always get filled on both sides evenly, so you end up holding a position you didn't want) and adverse selection (some of the people trading against you know something you don't, and you lose more on those fills than you make on the rest).

The three-line P&L

$\text{P\&L} = \underbrace{(\text{spread captured})}_{\text{revenue}} - \underbrace{(\text{inventory losses})}_{\text{holding cost}} - \underbrace{(\text{adverse selection})}_{\text{information cost}}$

Spread capture is simple: every round trip (buy then sell, in either order) earns roughly the quoted spread, minus fees. Inventory losses show up when the two sides don't balance — you get filled on ten buys and only six sells, and now you're long four units of something whose price can move against you before you unwind it. Adverse selection is the cost of being the easiest counterparty to trade against precisely when trading against you is profitable: informed sellers hit your bid right before the price drops.

Worked example: a full trading day, simplified

A market maker quotes a stock all day with a 2-cent spread (bid 99.99 / ask 100.01) and gets filled 10,000 times on each side — 20,000 fills total, fully balanced, so inventory ends flat.

• Gross spread capture: 10,000 round trips × $0.02 = $200.
• Fees: exchange charges $0.0005 per share on 20,000 shares = $10, roughly netted against maker rebates — assume net -$5.
• Adverse selection: markout studies (see Markouts: Measuring Post-Trade Drift) show that fills tend to be followed by a small unfavourable drift — say an average of $0.006 per share against the maker, on all 20,000 fills = -$120.

Net: $200 − $5 − $120 = $75 for the day, on 20,000 fills and $0.02 quoted spread. More than half the gross spread vanished into adverse selection alone — inventory losses are zero here only because the day happened to balance perfectly, which real days rarely do.

Why volume and turnover matter more than the spread itself

A wide spread earns more per fill but attracts fewer fills, since competing quotes step in front of you. A narrow spread earns less per fill but wins more of the flow — and if the flow is unbiased (roughly as many informed and uninformed trades either way), higher turnover on a thinner margin can beat lower turnover on a fat one, the same logic as a retailer choosing volume over markup. What breaks this is when a narrower spread doesn't just win more flow but wins worse flow — you're outcompeting other makers precisely on the fills they were smart enough to avoid.

Where this shows up in practice

• Quote width and skew are set jointly to manage the balance between capturing spread and avoiding toxic flow — see Skewing Quotes To Manage Inventory.
• Symbol selection. Names with wide spreads but heavy informed trading (illiquid small caps around news) are often worse economics than tight, boring, high-turnover large caps.
• Rebate capture. On maker-taker venues, exchange rebates for posting liquidity are a real part of revenue and change the breakeven spread — see P&L Attribution For A Market Maker for how it's split out.

In interviews

If asked "how does a market maker make money," resist the urge to stop at "the spread." The complete answer names all three terms — spread capture, inventory cost, adverse selection — and states plainly that the business is a bet the first outruns the other two, not a guarantee.