# How does crypto OTC liquidity work, and what do market makers do? An OTC quote draws on four sources: opposite client flow, the desk's inventory, market makers and exchange books. How makers price, hedge and why depth thins. --- Crypto OTC liquidity is the capacity of a desk and the firms behind it to absorb a large trade at one agreed price, and it comes from four places: an opposite client order, the desk's own inventory, external market makers and the order books of exchanges. A market maker is the firm that stands ready to buy and to sell at prices it sets itself, takes the client's position onto its own balance sheet, and then gets rid of the risk by hedging. The price you are shown is the cost of that chain: what it costs the market maker to hedge your size, what it charges for the risk it carries until the hedge is done, how much of the asset it already holds, and its margin. That is why doubling the size does not double the cost. Every price, size, volume and volatility figure in this article is an illustration built to show the mechanism. None is a statement of what desks or market makers quote in the market.
$7,800is what a bid 26 basis points below mid costs the seller of 1,000 units at a $3,000 reference
3.2xis how much the cost per unit rises when the size is multiplied by ten, under the square-root impact rule
$50,600is the modelled cost of a $2,000,000 order in a thinly traded token, against $1,900 in a deep market

Illustrative figures from the worked examples in this article; the square-root rule is an empirical regularity reported in academic research, applied here with assumed inputs.

## Where does the liquidity behind an OTC quote come from? The liquidity behind an OTC quote comes from four sources that a desk combines: another client who wants the opposite trade, the desk's own inventory and risk limits, prices streamed or quoted by external market makers, and the order books of exchanges. The client sees one price and one counterparty; the desk has usually assembled that price from several of these at once. Take a treasury that asks a desk to buy 4,000 ether from it, with a reference mid price of $3,000, a trade of $12,000,000. Mid means the point halfway between the best buying price and the best selling price in the market. The desk in this worked example fills the order from all four sources. | Source | Units | What limits it | | --- | --- | --- | | Opposite client interest, matched internally | 1,000 | Only exists if another client wants to buy at that moment | | The desk's own inventory, held as a position | 900 | The desk's risk limit and its capital | | External market makers, on request or by stream | 1,400 | Each maker's own inventory and hedging cost | | Exchange order books, worked over time | 700 | Visible depth, and the price moves as the order consumes it | | **Total** | **4,000** | | The internal match is the cheapest liquidity there is, since nobody has to carry market risk: one client's sale meets another client's purchase and the desk stands between them. It is also the least dependable, because it relies on coincidence. The desk's own inventory takes 900 units, a capacity set by its risk limit, which is a business decision and not a property of the market. External market makers supply the largest share, 1,400 units or 35% of the trade, and the remaining 700 units are sold on exchange order books in pieces that are small against the depth displayed. A desk that does not take risk itself, an agency desk or a broker, uses the same four sources without the second one: it finds the other side and passes the price through. How desks are organised and how they are paid are subjects of their own; what matters here is that a desk is a point of access to liquidity, and much of that liquidity belongs to someone else. ## What does a market maker do? A market maker quotes a price at which it will buy and a price at which it will sell, on a continuing basis, trades as principal against whoever accepts, and earns the difference between the two prices across many trades while trying to hold as little market risk as possible. Principal means the market maker is itself the buyer or the seller; it is not arranging a trade between two other parties. The United Kingdom's Financial Conduct Authority defines a market maker, in its Handbook glossary, as a person who holds themselves out as able and willing to enter into transactions of sale and purchase at prices determined by them generally and continuously, and not in respect of each particular transaction. A market maker is not a firm with a view on where the price is going. Its business is to be available on both sides. Three things follow from that role. - **It supplies immediacy.** A seller who wants to trade now does not have to wait for a natural buyer to appear. The market maker buys now and finds the buyer later. - **It warehouses risk for a short time.** Between buying from one party and selling to another, the market maker owns the position and the price can move against it. - **It needs two-sided flow to work.** A maker who only ever buys accumulates a position it has to pay to exit. In crypto markets the same firms often make markets in three places at once: on exchange order books, where they post the resting buy and sell orders that form the visible depth; in direct electronic streams to desks and large clients; and by answering individual price requests. The firm you face and the firm that ends up holding your risk are frequently not the same one. ## How does a market maker build the price it quotes? A market maker builds a quote by starting from a reference price and adjusting it for four things: the expected cost of hedging the size, a buffer for the price risk carried until the hedge is complete, a skew that reflects the inventory it already holds, and its margin. Each component can be expressed in basis points, where one basis point is one hundredth of one percent. The worked example below follows a market maker asked to bid for 1,000 ether from the treasury's order, with the reference mid at $3,000. The maker is already long 1,500 units and runs an inventory limit of 2,000.
How a market maker arrives at a bid for 1,000 unitsWorked example: reference mid $3,000, maker already long 1,500 units against a limit of 2,000
  1. Take the reference priceThe maker reads the mid across the venues where it can hedge: $3,000.00. Every adjustment below is subtracted from it, since the maker is buying.
  2. Estimate the cost of hedgingSelling 1,000 units back into the market will move the price. The maker's model puts that cost at 12 basis points, or $3.60 per unit.
  3. Add a buffer for the time the risk is heldThe hedge takes time and the market can fall meanwhile. The maker charges 5 basis points for that exposure, or $1.50 per unit.
  4. Skew for existing inventoryBuying 1,000 units would take the position to 2,500, which is 500 over the limit. To be paid for holding more of what it already has too much of, the maker lowers its bid by 6 basis points, or $1.80 per unit.
  5. Add the marginThe maker's intended earning on the trade is 3 basis points, or $0.90 per unit.
  6. QuoteThe total is 26 basis points, or $7.80 per unit. The bid is $2,992.20, and the client selling 1,000 units receives $2,992,200, which is $7,800 less than the mid value of $3,000,000.
Only the last of the four components is what the maker intends to keep. The hedging cost is paid away to the market, the risk buffer is a reserve that is sometimes kept and sometimes lost, and the skew is a statement about the maker's own book. Reading the whole 26 basis points as the maker's profit misreads the quote. The hedging cost is broadly similar for every maker, since all hedge in the same markets. The inventory skew is not: a maker who happens to be short the asset wants to buy it, and will pay closer to mid to get it. ## Why can one market maker show a poor bid and a good offer at the same moment? A market maker shows a poor bid and a good offer at the same moment when it already holds more of the asset than it wants: it lowers both of its prices so that it is less likely to buy more and more likely to sell. The quote is centred on the price at which the maker is indifferent, and that price is not the market mid. In the worked example, the maker who bids $2,992.20 for 1,000 units is, at the same moment, willing to sell 1,000 units at $3,000.60, only 2 basis points above mid. A sale reduces its inventory from 1,500 to 500 and needs no hedge, so the hedging cost falls away and the skew works in the buyer's favour: 3 points of margin and 5 of buffer, less 6 of skew. The two-sided quote is therefore $2,992.20 bid and $3,000.60 offered, a width of $8.40 or 28 basis points, centred on $2,996.40, which sits 12 basis points below the mid. This behaviour has a standard description in the academic literature. In the model published by Marco Avellaneda and Sasha Stoikov in the journal Quantitative Finance in 2008, a dealer sets quotes around a "reservation price" that is the mid adjusted for its inventory, and when the dealer is long, that reservation price lies below the mid, which expresses a wish to sell. For a client, a bid well below those of other firms does not necessarily mean the firm is expensive; it may mean the firm is full. The same firm may be the most competitive on the day you want to buy. ## How does a market maker hedge a trade it has just taken? A market maker hedges a trade by taking an offsetting position as fast as the market allows: selling the asset on exchanges, selling a futures or perpetual contract on the same asset, trading with another desk, or holding the position until an opposite client order arrives. The routes differ in speed and in what they leave behind. - **Spot sales on exchanges** remove the position entirely, but each sale consumes visible depth, so a large hedge is spread over time. - **A derivative hedge**, where a liquid futures or perpetual market exists for the asset, removes the price risk within seconds while the maker still holds the units. The spot position is then sold gradually, and the maker bears the risk that the two prices drift apart. - **A trade with another desk or maker** passes the risk on in one piece, at that firm's price. - **Waiting for opposite client flow** costs nothing in the market and leaves the maker fully exposed until the flow arrives. The risk buffer in the quote exists because hedging is not instantaneous. Suppose the maker in the example buys 1,000 units at $2,992.20 and sells all of them at the price its model expected, 12 basis points below the mid of $3,000, which is $2,996.40. It earns $4.20 per unit, or $4,200, equal to 14 basis points on the trade: its 3-point margin, the 5-point buffer it did not need and the 6-point skew. Now suppose it sells only 600 units at $2,996.40 before the market falls by 0.5%, taking the mid to $2,985.00. The remaining 400 units are sold 12 basis points below the new mid, at $2,981.418. The first 600 units earn $2,520. The last 400 lose $10.782 each, or $4,312.80. The trade ends $1,792.80 in the red, after a market move of half a percent. A maker earns a few basis points when a trade goes to plan and can lose several times that when the market moves during the hedge. ## Why does liquidity thin as the size of the trade grows? Liquidity thins as size grows because every source behind the quote is finite: the order book holds a limited quantity near the current price, each market maker has an inventory limit, and the hedging of a large trade pushes the price against the firm doing it. The cost per unit therefore rises with size, although less than proportionally. Academic studies of trade data describe the shape of that rise with what is called the square-root impact law. As stated by Bence Tóth, Jean-Philippe Bouchaud and co-authors in work published in Physical Review X in 2011 and in later papers, the average price impact of a large order is approximately the asset's daily volatility multiplied by the square root of the order's size as a share of daily traded volume, times a constant of order one. The same authors report that the relation has been observed on futures, on stocks and on bitcoin. The chart applies that rule with assumed inputs: a constant of one and a daily volatility of 3%.
Modelled price impact as the order growsBasis points, square-root rule with a constant of 1 and a daily volatility of 3%; size as a share of the asset's daily traded volume
0.1% of daily volume 9.5 bps 1% of daily volume 30.0 bps 5% of daily volume 67.1 bps 10% of daily volume 94.9 bps 25% of daily volume 150.0 bps

Illustration of the shape of the relation, not a forecast for any asset. The constant and the volatility are assumptions; real impact varies by asset, venue and market conditions.

Two readings of the chart matter for a client. First, size is measured against the market, not in dollars: the relevant number is the share of a day's volume the order represents. Second, the curve bends. An order 250 times larger, from 0.1% to 25% of daily volume, costs about 16 times more per unit, not 250 times. The rule describes hedging in the open market. A desk that can match the order internally, or a maker that wants the position, can quote inside it. ## Why is liquidity thinner in small tokens? Liquidity is thinner in small tokens because fewer firms make markets in them, daily traded volume is low so any order is a large share of it, volatility is higher, and there is often no derivative market in which a maker can hedge quickly. Each of those four conditions widens one of the components of the quote. The table applies the same square-root rule to one order of $2,000,000 in three situations. The inputs are assumptions. | Situation | Daily volume and volatility | Order as share of volume | Modelled cost | | --- | --- | --- | --- | | Large asset, calm market | $2,000,000,000 and 3% | 0.1% | 9.5 bps, or $1,900 | | Large asset, stressed market | $2,000,000,000 and 9% | 0.1% | 28.5 bps, or $5,700 | | Small token, calm market | $20,000,000 and 8% | 10% | 253.0 bps, or $50,600 | The same dollar amount costs about 27 times more in the small token than in the large asset. Volume does most of the work: the order is 10% of a day's trading instead of 0.1%. The model understates the difficulty in two ways that a desk will raise with you. A maker cannot easily go short a token it cannot borrow. A client who wants to buy a large amount of a small token is asking the maker to find sellers, and if the supply sits with a few holders, there may be no quick way to source it at any price. The quote then comes with a longer execution period, or with no guaranteed price. Market making in small tokens is frequently arranged by the token's issuer under a contract with one or two firms. This is ordinary practice, and it concentrates liquidity in very few hands. The International Organization of Securities Commissions, in its 2023 policy recommendations for crypto and digital asset markets, asks regulators to require service providers to disclose whether they or their affiliates engage in market-making and whether client trades are made with them on a principal basis, and it singles out the conflict that arises when a firm involved in issuing or listing a token also makes the market in it. ## What happens to OTC liquidity when the market is under stress? In a stressed market OTC liquidity contracts on every dimension at once: quotes widen, the size a maker will price falls, quotes stay valid for a shorter time, and some assets stop being quoted. The mechanism follows from the quote components. Volatility rises, so the risk buffer grows, as the second row of the table above shows: three times the volatility gives three times the modelled cost for the same order. Order books thin out because the makers who post them widen or reduce their own orders, which raises the hedging cost. Client flow becomes one-sided, since most clients want to sell at the same time, so internal matches become rare and every maker's inventory fills in the same direction. The skew that was a matter of chance in calm markets now points the same way at every firm. This pattern is not specific to crypto. The Committee on the Global Financial System at the Bank for International Settlements, in its 2014 study of market-making and proprietary trading, reported signs of "liquidity bifurcation and fragility", with market activity concentrating in the most liquid instruments and deteriorating in the less liquid ones. Stress also reaches credit. Market makers trade with one another and with desks on credit lines and settlement limits. When firms grow cautious about each other, they cut those limits, and the amount of trading the same capital can support shrinks. ## What does thin liquidity mean for the price you receive? Thin liquidity means the price you receive moves away from the reference price by an amount set mainly by three things you can estimate beforehand: your size as a share of daily volume, the asset's volatility, and whether the maker you face wants your side of the trade. The first two set the general level of the quote; the third explains the differences between firms. On the treasury's full order of 4,000 units, a cost of 26 basis points against a mid value of $12,000,000 would be $31,200. Whether that is a fair price cannot be judged against zero. It has to be judged against what the same sale would cost by another route: the impact of working it through the order books, and the risk carried while doing so. Two practical consequences follow. A quote for the whole size is not the only form of liquidity. A desk can price part of the order firmly now and work the remainder, and the blended result can beat a single price for everything when the last units are the expensive ones. Information is part of the cost. A maker who learns that a large sale is coming, and then does not win the trade, still knows it is coming. Showing the full size to many firms at once can move the market before you trade, which is why large orders are often shown to a small number of firms. ## What does thin liquidity mean for your timing? Thin liquidity makes timing a real choice between two costs: trading immediately transfers the risk to the maker and you pay for it in the price, while trading over hours or days lowers the market impact and leaves you carrying the price risk meanwhile. Neither is free, and the right balance depends on how much price movement you can tolerate. The table shows the second cost for the treasury's $12,000,000 position, using the same 3% daily volatility and the usual scaling of volatility with the square root of time. A "typical move" here is one standard deviation, in either direction. | Execution window | Typical price move over the window | On $12,000,000 | | --- | --- | --- | | 1 day | 3.0% | $360,000 | | 3 days | 5.2% | $624,000 | | 5 days | 6.7% | $804,000 | Against a one-off cost of $31,200 for immediate risk transfer, an exposure of several hundred thousand dollars in either direction is a different order of magnitude. The move can go in your favour as easily as against you, so this is not a cost in the way a spread is. It is uncertainty, and a treasury with a budget to meet may reasonably pay to remove it. A holder with no deadline, selling a small token quoted several percent away from mid, may reasonably choose the opposite. Within a day, liquidity also varies by the hour. Crypto markets trade continuously, but the firms that make them are staffed in particular time zones and banking hours govern when fiat can move, so depth is not the same at every hour or on every day of the week. Scheduled announcements and the minutes around them are poor moments to ask for a large price. ## What should you ask a desk about its liquidity? Ask a desk where the liquidity behind its quote comes from, in what capacity it is trading with you, what size and time its price is good for, and how it will handle the part of the order it cannot price at once. A desk that deals in size answers these questions routinely. - Will you trade with me as principal, or arrange the trade as agent? - Do you make markets yourself in this asset, rely on external market makers, or both? How many makers do you see in it? - What size is this price good for, and for how long? - If my full size cannot be priced firmly, what do you propose for the remainder, and how will its price be set and reported? - For a small token: who makes the market in it, and what depth do you see on the venues you can use? - How many firms will learn of my order, and will they see its direction? A desk that says it will source most of a small-token order from a single maker over two days has told you what the liquidity is. The settlement of the trade, meaning who delivers first and how each side is protected until both legs are complete, is a separate question to agree in writing once the price is done.