The Market Making Book

3. What Is a Market Maker?

A merchant of immediacy: always willing to buy, always willing to sell, and paid the spread for never saying no.

Part I · Chapter 3

A market maker (MM) simultaneously posts a bid and an ask in the same instrument, hoping to buy at the bid, sell at the ask, and capture the spread — over and over. The MM doesn't try to predict where the price is going; it tries to be the counterparty to everyone else's impatience.

Parable · The fruit vendor Rosa runs a fruit stall. She buys apples from anyone at $0.95 and sells to anyone at $1.05. On a good day, farmers and snackers arrive in roughly equal numbers and she earns 10¢ per round trip, hundreds of times. But two things keep Rosa up at night. One: some afternoons only farmers show up — she ends the day buried in apples, and if tomorrow's apple price drops, her pile loses value. Two: sometimes the person selling her apples is an agronomist who knows a record harvest will crash prices tomorrow. Rosa's two fears have formal names: inventory risk and adverse selection. Every formula in this book is a way of pricing one of Rosa's fears.

Watch Rosa's business below as a simulation: a mid-price that drifts randomly, quotes around it, traders arriving at random and hitting either side.

simulation — earning the spread
Round trips0
Inventory q0
Spread P&L0.00
Inventory P&L0.00
Total0.00
Tighten the spread: more fills, smaller margin per fill. Raise volatility: inventory P&L starts to swing your total around — the spread income is steady, but the value of what you're holding is not. That swing is inventory risk made visible.

The two P&L streams

The figure splits the MM's profit into the two streams every desk monitors separately:

  • Spread P&L — the accumulated half-spreads from every fill. Steady, almost mechanical income. This is the business.
  • Inventory P&L — the mark-to-market change of whatever you happen to be holding. Pure noise at best, a slow bleed at worst. This is the risk.

A good market making operation maximizes the first stream while strangling the variance of the second. Notice in the simulation that inventory drifts away from zero on its own — random fills don't balance perfectly. Left unmanaged, the position grows like a drunkard's walk, and with it your exposure. Chapter 6 derives exactly how much you should shade your quotes to push inventory back toward zero.

The economics of one round trip

It pays to write the business down as an accountant would. One round trip — buy at the bid, later sell at the ask — earns the spread, minus everything that happened in between:

P&L of one round trip E[P&L] = spread  −  2·fees  −  E[adverse move while holding]  −  hedging costs What the symbols mean E[·] — "expected value": the average over many repetitions, not what any single trade does  ·  spread — what you pocket by buying at your bid and selling at your ask  ·  2·fees — the venue charges you twice, once per side of the trip  ·  adverse move — how much the price drifts against you, on average, while you're stuck holding the position  ·  hedging costs — what you pay (fees + crossing the spread elsewhere) if you offload the risk on another venue.

Each negative term gets its own chapter: fees are venue terrain (Part III), the adverse move splits into inventory risk and adverse selection (Chapter 5), and the time you spend holding — which scales the adverse-move term — is exactly what the mathematics of Chapter 6 minimizes. Two further subtleties hide in the word "round trip":

  • Round trips require both sides to fill. In a balanced, mean-reverting market they do, constantly. In a trending market only one side fills — there is no round trip, just a growing position. The trend doesn't reduce your income; it converts your business from spread-collection into an unintended directional bet.
  • The halves of the trip are not independent. The fill that opens your position is more likely to arrive precisely when the price is about to move against you (Chapter 5 makes this rigorous). Averaging "spread per round trip" over good days hides the bad fills that never closed.
interactive — the round-trip waterfall
Net per round trip
Verdict
The whole business on one chart: the spread builds the only green bar, and every cost takes its bite. Now run the venue thought-experiments: set fees to a Kalshi taker's (~1.75¢ per side at 50¢ — the green bar drowns instantly: that's why Chapter 9 says maker-only); crank the adverse move to a courtsider's jump and watch no realistic spread survive. Profitable market making is not finding a magic spread — it's keeping three red bars short.

Why someone pays you the spread

Takers cross the spread because immediacy is worth more to them than the spread costs: a hedger needs the position now, a retail trader doesn't care about 1¢, an arbitrageur is locking a larger profit elsewhere. As long as a healthy share of the flow trades for reasons other than superior information, the kiosk business works. The moment most of your counterparties know something you don't, you are no longer a kiosk — you are the sucker at the table. That's Chapter 5.

Definition worth memorizingMarket making = short-volatility-flavored income (the spread) funded by providing immediacy, taxed by two costs: the variance of your inventory and the information of your counterparties.

On this page

GitHubGitHub repository