Market makers

The job: always be willing to trade

A market maker is a firm that posts a price to buy (the bid) and a price to sell (the ask) at the same time, all day, and stands ready to trade with whoever shows up. In the order book from chapter 2, many of the resting orders at the inside of the book belong to market makers. They are not betting on whether the stock goes up. They are trying to buy at the bid, sell at the ask a few seconds or minutes later, and pocket the difference.

That difference is the spread. If a maker quotes $49.99 bid and $50.01 ask, and one person sells to them while another buys from them, they have bought a share for $49.99 and sold it for $50.01. Two cents a share is not much. Do it on hundreds of millions of shares a day across thousands of stocks and it becomes a business.

The service they sell is immediacy. Without them, a seller would have to wait until a buyer happened to arrive, or keep lowering the price until one did. The spread is what you pay to not wait. In heavily traded US stocks it is often a single cent, which for most stocks priced above a dollar is the smallest price step the rules allow. In thinly traded stocks it can be many cents, or a percent or more of the price.

Three ways a market maker loses money

If earning the spread were the whole story, everyone would do it and spreads would shrink to nothing. They don't, because the spread has to pay for three kinds of risk.

1. Inventory risk

Order flow does not arrive neatly in pairs. Ten sellers in a row might hit the bid before a single buyer lifts the ask. Now the maker owns a pile of shares they never wanted, and if the price drops before they can sell them, the loss on that pile can wipe out a day of spreads. The longer they hold a position and the more volatile the stock, the bigger this risk.

The standard defence is to skew the quotes. A maker who is long (owns too many shares) lowers both their bid and their ask a little. The lower ask makes buyers more likely to pick them, which sells inventory off; the lower bid makes sellers less likely to pick them, which stops the pile growing. A maker who is short skews the other way. Skewing costs something, because quotes that lean away from the middle earn less spread, but it keeps the position small.

2. Adverse selection

This is the one that defines the business. Some of the people trading with a market maker know something the maker doesn't: a fund that has just finished a week of research, a trader who saw a news headline half a second earlier, an algorithm that noticed the same stock moving on another exchange. When an informed trader buys from you, it is because the price is about to go up. You sold at $50.01 and the stock is worth $50.20. You earned a cent of spread and lost twenty cents of value.

The maker cannot tell informed from uninformed traders in advance; the order just says buy 300 shares. So they set the spread wide enough that what they earn from the uninformed (people investing their paycheck, funds rebalancing, someone selling to pay for a house) covers what they lose to the informed. Economists call this the adverse selection component of the spread, and a well-known model of it (Glosten and Milgrom, 1985) shows a spread can exist even if quoting is free and makers compete hard, purely because some traders know more.

Two practical consequences follow. Spreads widen when information risk is high: just before earnings, around big economic releases, in small companies few analysts follow. And a maker learns from the flow itself: if buyers keep lifting the ask, maybe they know something, so the maker raises their quotes. That is one of the main ways information gets into prices, which chapter 6 picks up.

3. Being picked off when the world moves

A quote is a promise that sits in the book until it is cancelled. When news breaks, there is a race between the maker cancelling a stale quote and someone trading against it. Firms spend heavily on fast connections and fast code mostly to win this race, so their quotes are not sitting there at yesterday's price when the world changes. Volatile markets make stale quotes more dangerous, so makers widen spreads and post smaller sizes when prices are jumping around.

Reading your desk

The profit is split the way real trading desks split it.

• Spread earned is how far each trade was from the mid at the moment it happened: buying 100 shares two cents under the mid earns $2. This is the income.
• Lost to informed flow is a markout: for each trade, how far the mid moved against you over the next few minutes. Buy 100 shares and see the mid fall 10 cents shortly after, and that trade marks out at minus $10. Real desks track markouts obsessively, broken down by who sent the order, because they are the best measure of how informed a counterparty is.
• Inventory held longer is everything else: the gain or loss on shares you were still holding after that window, which is mostly luck, good or bad, and grows with the size of your position.

A few experiments worth running:

• Turn informed traders up to 30% and leave everything else alone. Spread income barely changes, but the markout line dives. This is a stock with an information problem, and the only answer is a wider spread.
• Now widen your half-spread to 8 cents. You might expect that to fix it. Watch the trade count collapse instead: the other dealers quote 3 cents either side and take nearly all the uninformed traders, but the informed traders still come to you whenever the news makes your stale quote a bargain. You are left with mostly the worst flow. This is why makers can't simply quote wide and relax.
• Set skew to zero. Your inventory wanders to many thousands of shares, and the last line on the chart swings wildly from day to day. Press New day a few times and compare.
• Set skew to the maximum. Inventory stays tiny, but your quotes lean so hard that you sometimes buy above the mid. Spread income falls. Real makers tune this constantly.

Who the market makers are

The term covers several different businesses.

Designated market makers have formal obligations to an exchange. At the New York Stock Exchange, each listed stock has a designated market maker that must help keep the market orderly, run the opening and closing auctions with the exchange, and quote within certain ranges, in return for some advantages. This is the modern descendant of the old floor specialist.

Electronic market makers are trading firms that quote on many exchanges at once using automated systems, often described as high-frequency trading firms. Exchanges register many of them as market makers, with lighter obligations to keep a quote up. Most of the visible displayed liquidity in large US stocks comes from firms like these, and they are also major market makers in ETFs, options and futures.

Wholesalers are market makers that trade directly with orders sent by retail brokers, away from the exchanges. When you buy shares in a zero-commission app, your order is often filled by a wholesaler rather than on an exchange, and the broker may be paid for sending it there. That arrangement, payment for order flow, gets its own section in chapter 5. The short version of why it exists is in this chapter: retail orders are mostly uninformed, so trading against them is low-risk, and wholesalers compete to fill them, often at a price slightly better than the best displayed quote.

Useful, and criticised

The case for market makers is the spread itself. Over the past few decades, spreads in US stocks have shrunk enormously, helped by decimal pricing (prices moved from fractions like 1/16 of a dollar to pennies in 2001) and by competition between electronic firms. For an ordinary investor buying an index fund or a large company's shares, the cost of trading is now a small fraction of what it was.

The criticisms are real too:

• Liquidity can vanish when it is most needed. Most electronic makers have no obligation to keep quoting in a crash. When the risk of being picked off soars, the rational response is to widen spreads or step away, which is exactly when sellers most need a buyer. Chapter 6 shows what that does to a price.
• Speed advantages. Firms that see and react to prices microseconds faster can avoid being picked off, and can sometimes pick off slower traders. Whether that is a cost to ordinary investors, or mostly a fight between professionals, is still argued about.
• Conflicts in the retail pipeline. When a broker is paid by the firm that fills its customers' orders, the customer has to trust that the broker is still getting them the best price. Regulators require it, and the fills are usually at or better than the quoted price, but the incentive is there.

Next: the path your order actually takes from the app on your phone to whoever fills it, and what happens in the day after the trade.