Trading

Market Making

Market makers have come to dominate a region of finance, and generate huge profits and compensation for their employees. In this section, we outline the concept of market making and how market making firms generate profit.

Introduction Markets Limit-Order Book Arbitrage Market Making Signal-Based HFT Latency Statistical Arbitrage

Another common type of trading strategy is market making. This strategy was originally carried out in the 'open outcry' style, in trading pits, and the automation of this strategy directly led to the birth of many modern, successful high-frequency trading firms, such as Optiver and Susquehanna. Proponents of this strategy place limit orders passively — bids below the current best offer, and offers above the current best bid — in an effort to trade against price-insensitive market participants who cross the bid-offer spread. Since this strategy relies on waiting for others to trade, it is qualitatively different from an arbitrage strategy.

A graph illustrating the limit-order book. The x-axis is the asset price, and the y-axis is the quantity of the asset offered at that price. Bids are shown in green, asks are shown in red. The gap between bids and asks is known as the bid-ask spread.

A graph illustrating the limit-order book.

For example, here we could place a bid at the best bid and an offer at the best offer. If things go well, a seller will arrive in the market, and a buyer at around the same time, and we will buy at the best bid and sell at the best offer, making a small profit equal to the bid-offer spread.

It is irrational to place an order at a price worse than you think is fair for that product; in other words, all of our orders need positive edge, which is the expected profit of the order, measured against its fair price:
$$\text{edge} = \begin{cases} \text{fair price} - \text{order price}, & \text{for bids} \\ \text{order price} - \text{fair price}, & \text{for offers} \end{cases}$$

In particular, say we think the fair price of the above product is the mid-price. We could then also place a bid one tick above the best bid and an offer one tick below the best offer, and our orders would still have edge. These orders would have less edge than the first set, but better price priority, making them more likely to trade quickly and generate a profit.

Let's take a step back to see what has happened here. By providing orders for others to trade against around our fair price:

  • We have contributed to the consensus mechanism that establishes this product's fair price.
  • With the second set of orders, we have actively improved the best price at which transactors can trade.
  • Even with just the first set, we have added book depth: if somebody wishes to transact a large size, they will match against more quantity at a better price, improving the average price of their transaction.

Through all of these methods, market makers contribute to the efficient functioning of the markets.

A market maker estimates that a product has a fair price of 100. The tick size is 0.1. At what price should she place her bid in each of the following two markets?

Her bid price must be strictly lower than 100, or she will have no edge. In fact, a bid at 100 would not only be unprofitable — it would actually lose her money on average. This is because any fair-price estimate carries uncertainty, and by bidding 100 she would trade more often when her estimate is wrong than when it is right — for example, against the arbitrageurs of the previous section. Other considerations, such as exchange fees and taxes, require her to take more edge on each trade to break even. However, since the market's best bid is already at 100, after these considerations she should place her bid as high as possible — i.e. 99.9 — to maximise her chances of trading in the first place.

Even after the above considerations, she may bid higher than 99: 1% of edge is relatively high in a market-making context, and should be more than enough to remain profitable even after accounting for her model uncertainty, exchange fees and taxes. But should she ever bid higher than 99.1? Doing so only decreases her edge on each trade, whereas bidding 99.1 is already sufficient to maximise her chances of trading against a transactor crossing the market. So she should bid 99.1.