A dive into liquidity demographics for crypto asset trading

In the fast-evolving and often volatile landscape of cryptocurrency markets, liquidity plays a pivotal role in determining efficiency, stability, and resilience. In the crypto markets, much like traditional finance, liquidity measures how quickly an asset can be transformed into fiat or stablecoins without significant costs or price dislocations. Small transaction costs, timely trading, clearing and settlement, and limited price impact are typical characteristics of liquid financial assets. While liquidity in crypto asset trading exhibits similar characteristics to that of traditional assets, there are also key differences that make the crypto ecosystem unique, as we will discuss in this report.

To assess the liquidity profile of a digital asset, we look at: 

• Market volume: quantifiable by the number of transactions and volume (expressed in a fiat currency). 

• Bid-ask spread: calculated by subtracting the bid, or the highest price a buyer is willing to pay, from the ask, or the lowest price a seller is willing to accept.

• Market depth: measurable by the fiat equivalent potential trades within a range of the current price.

• Slippage: quantifiable by the difference between the expected and actual outcome of a trade, measured by the number of tokens exchanged.

For the traditional financial industry, trading hours are an important dimension of liquidity. By contrast, crypto markets are open globally 24/7, unbound by banking hours or intermediaries. The execution speed for crypto markets is related more to technology than to a third party’s performance. Traditional finance exchanges rely on intermediaries such as banks to clear and execute transactions, while decentralized finance (DeFi) relies on blockchain technology, utilizing a decentralized network that functions based on smart contracts to process transactions. 

In this report, we perform a liquidity analysis on crypto assets trading on centralized exchanges and on decentralized exchanges. We will look at some large exchanges and study liquidity profiles on selected markets. The report covers digital assets that are issued or transferred using blockchain technology, such as cryptocurrencies and stablecoins.

In the remainder of the report, we calculate liquidity metrics for selected markets that include three stablecoins (USDC, USDT, DAI) and two cryptocurrencies (BTC and ETH). We looked at the historical data from Jan 1, 2023, to Feb. 28, 2025. For CEXs, we use data from Binance, and for DEXs, we use data from Uniswap V3. Each of the pairs we analyze represents a standalone segment of the crypto market that trades on a specific exchange and exhibits certain liquidity characteristics. 

The launch of Bitcoin (BTC) and Ether (ETH) exchange-traded funds (ETFs) has been a significant event in the crypto market. ETFs allow traditional investors to gain exposure to BTC and ETH without directly purchasing the underlying assets. These products have the potential to increase liquidity, impact volatility, and alter market dynamics by attracting institutional investors. 

Figure 4 shows that BTC and ETH trading volume increased on Binance around the launch on NASDAQ and NYSE of the IBIT and Grayscale ETH ETFs. 

We present below four liquidity metrics: volume, bid-ask spread, market depth and slippage.

Trading volume is a transaction-based metric that indicates the level of activity and liquidity for an asset by combining the number of trades and the execution price of each trade in a certain time interval. In the subsequent graphs, volume includes the US dollar volume for all markets on a given exchange, fiat or otherwise. For example, on Binance, the US dollar value of the daily volumes for all markets with fiat currencies and crypto assets paired with BTC will be counted towards the total daily volume for BTC. 

As of this analysis, crypto markets’ liquidity remains fragmented. Trading volume is spread out across multiple different platforms and exchanges, and this can lead to price differences, market inefficiencies, and execution challenges for larger trades. 

Figures 6 through 10 show the average daily volume in US dollars from Jan. 3, 2023, to Feb. 28, 2025, over various exchanges, expressed as percentages of the total volume listed on these exchanges. See the appendix, section three, for tables showing the mean daily volume of crypto assets on various exchanges.

Besides the fragmentation of the crypto markets, political events and cyber hacks have a direct impact on price and liquidity. At the beginning of December 2024, the price of Bitcoin (BTC) on the KRW (Korean Won) market on Upbit exchange saw a significant drop due to a local political crisis in South Korea. This event highlights that liquidity can differ across exchanges and markets. Localized events can cause price discrepancies even when the global price is relatively stable. Figure 11 shows that on Dec. 3, 2024, while the global BTC price declined modestly to $94,000 (KRW 132 million, data source: Kaiko) on Upbit, the volume weighted average price for BTC-KRW  plummeted to as low as  $63,786 (KRW 89.84 million, data source: Kaiko) between 1:30 PM and 2:59 PM UTC.

On Feb. 21, 2025, Bybit, one of the world's largest cryptocurrency exchanges, experienced a security breach, resulting in the theft of about $1.4 billion worth of Ether (ETH) and related tokens from its cold wallet. Our analysis of trading volume before and after that day found a significant volume spike on the day of the hack followed by a lowered volume regime (about half of pre-attack) in the immediate aftermath of the event (see Digital Assets Brief: Bybit Hack Underlines Importance Of Cyber Resilience). The stolen funds were part of the exchange’s reserve, and the daily volume in Figure 12 does not include the illicit transfer.

Bid-ask spread is an order book-based measure, based on executable trade quotes. The narrower the spread, the more liquid the market is for that asset. Figure 13 shows the time series of the daily bid-ask spread for selected stocks and digital assets-fiat markets. For each order book, we calculated the difference between the lowest ask and the highest bid, normalized by the midpoint price. We then calculated a daily weighted bid-ask spread at the asset level.

In this paper, we compared the historical bid-ask spreads of two examples of very liquid large-cap stocks in the S&P 500, Apple and Broadcom, with those of selected digital assets. Figure 13 shows that on Binance, BTC, ETH and USDT typically have a higher bid-ask spread than Apple (AAPL), reflecting lower liquidity and potentially higher trading costs. 

Figure 14 shows the opposite trend when comparing the selected digital assets (BTC, ETH, USDT and USDC) to Broadcom (AVGO). In this case, the digital assets have tighter bid-ask spreads. 

Figure 15 shows the historical average for the weighted bid-ask spread for Broadcom is higher than that of the crypto assets, while Apple’s bid-ask spreads are lower.

We next looked at the correlation between the bid-ask spreads of the various assets. Figure 16 shows high bid-ask correlation between BTC and ETH (0.72), as they are being influenced by the same market factors, such as market sentiment, trading activity and overall volatility. 

The stablecoins exhibit low correlation with the cryptocurrencies, which is somewhat expected. Also not surprising is the low correlation between crypto assets and stocks, as the drivers for cryptocurrency valuation and liquidity are different from those of traditional financial data. The two stocks are profiling different sectors (tech and semiconductors) and exhibit insignificant bid-ask correlation as well.

Market depth is an order-book based metric that measures the exchange’s capacity for large orders without impacting the price of the asset more than a given percentage. It speaks to the degree to which an asset can be bought or sold without causing price instability. For example, a total 1% depth measures the volume of executable trades within a 1% range of the quoted mid-price of the asset. If the market depth is shallow, it is easy for a large trade to significantly impact the price. As a result of fragmented liquidity, it is not unusual to observe large deviations in prices on certain exchanges. Given that the crypto market is fragmented, we calculate the depth for specific markets on Binance. Of note is the fact that the market depth could vary significantly on other exchanges and smaller markets.

Figure 17 shows the market depth on Binance for the asset pair EUR-USDT in euros for the 1% and 10% band (up and down) from the midpoint of the bid-ask price. Of note is the shallow market of less than €5 million for the 1% band and less than €8 million for the cumulative 10% band. This is not surprising as stablecoins have more utility on-chain than off-chain. 

By contrast, on Binance, USDT liquidity is much greater against other digital assets such as ETH and BTC than against fiat markets. Figure 18 shows a much higher market depth for BTC-USDT and ETH-USDT for the 1% band around the mid-price of USDT 15 million and USDT 30 million, respectively.

Figure 19 shows the average 1% market depth for selected markets (Binance) is lowest for EUR-USDC and highest for BTC-USDT. Crypto-USDT markets show a significantly higher market depth than crypto-fiat markets, which involve banks, fees, know your customer and other compliance measures. Crypto-USDT does not have any traditional banking constraints and is integrated seamlessly into trading strategies. 

External events or market news can affect market depth. During events of high volatility, such as the Silicon Valley Bank crisis in March 2023, the order book becomes less stable, and price movements can be erratic. Figure 20 shows the widening of the gap between market depths at 1% and 10% levels on Kraken for USDC-USD.

There were few executable trades at the 1% level near the peg, while at the 10% level, market participants might have been more willing to buy larger orders, leading to more liquidity and widening the gap between the 1% and 10% levels. 

Next, we looked at a cross-sectional analysis, which gives us a view of the entire order book up to 10% up and down from the mid-price for a given date (Feb. 28, 2025). This allows us to compare the liquidity profile of various crypto assets at a given point in time. Figure 21 shows the cumulative depth (executable trades) of BTC, ETH, USDC and USDT against fiat EUR at various percentages from the mid-price. 

For example, the Bid Volume 10% of €3.2 million for BTC on Binance represents the euro volume of all bids falling below the mid-price (see blue bars) within a 10% price range. Conversely, Ask Volume 10% of €1 million for BTC on Binance represents the EUR volume of all asks falling above the mid-price (see red bars) within a 10% price range on that day. 

Slippage quantifies the difference between the expected and actual outcome of a trade. For crypto asset trading on decentralized exchanges, it is calculated as: (expected number of tokens-actual number of tokens received)/expected number of tokens.

Because the number of tokens exchanged implies a price of one token in units of measure of the other token, slippage also measures the percentage difference between the expected price of a trade and the actual price at which it is executed. 

A large trade (with respect to the pool size) can cause significant deviation between the executed price and the expected price of a trade. Slippage provides information on how much a price for a trade can change based on its size. High market depth and low slippage are indicators of a liquid market. We examine two liquidity pools for two trading pairs on Uniswap V3 on Ethereum:

1. USDC-ETH pair (0x88e6A0c2dDD26FEEb64F039a2c41296FcB3f5640), which had a total pool size of $158.8 million as of March 24, 2025. 
2. DAI-USDC pair (0x5777d92f208679DB4b9778590Fa3CAB3aC9e2168), which had a total pool size of $62.0 million as of March 24, 2025.

To illustrate slippage, we used a hypothetical trade size of one million tokens, and we used hourly sampling in a 24-hour period from April 2023 to October 2024. We plotted the maximum and minimum slippage values over consecutive 24-hour periods for the sampled data. Maximum slippage over a 24-hour period highlights the worst-case price impact for large trades during a day. Trends in slippage over time can explain the volatility of trading pairs. 

Below, we analyze the slippage when selling USDC to buy ETH or DAI. Potential use cases for this scenario include: 

• Loan repayment: If one needs ETH to repay a loan, a trader may sell USDC to get ETH.

• Interest rate arbitrage: If lending DAI earns 7% and borrowing USDC costs 6%, a trader may sell USDC to acquire more DAI.

Figure 22 shows the maximum and minimum slippage over a 24-hour period for selling one million tokens of USDC to buy ETH for our dataset.

On Feb 3, 2025, we see a minimum slippage of 0.05% and maximum slippage of 4.96% for the sampled data. Note that the maximum and minimum slippage occur for different hypothetical trades and different times. ETH price dropped by 18.45% between Feb. 1 and Feb. 3 during the maximum slippage period depicted in the above graph.

To interpret the slippage, let’s assume a hypothetical scenario in which 1 USDC was worth 0.01 ETH. When the trade was initiated to sell one million USDC, the participant expected to get 10,000 ETH. However, they got 9,504 ETH at the time of execution due to the slippage of 4.96%. Similarly, for the minimum slippage of 0.05%, the buyer got 9,995 ETH when the trade was executed. 

Figure 23 shows the maximum and minimum slippage over a 24-hour period for selling USDC to buy DAI. The basis between maxima and minima is exceedingly small, in the order of 10^ (-6).

On March 22, 2024, we see a maximum slippage of 0.0102734%. The very low slippage could be attributed to primary redeemability and/or a peg stability module. 

The liquidity of crypto markets is generally lower than traditional financial markets, it is fragmented, and it is evolving. While liquidity has been improving over the years with the development of more robust platforms, financial instruments (such as ETFs), and the entry of institutional investors, challenges still exist. Liquidity demographics depend on specific markets and asset pairs (fiat-crypto or stablecoin-crypto, crypto-crypto). Cryptocurrency liquidity will evolve, influenced by regulatory developments, technological advancements and increasing institutional participation. Supportive regulatory frameworks and institutional involvement could create a more robust and efficient crypto ecosystem. 

Centralized crypto-trading platforms (CEX) rely on a central limit order book operated on a system owned by the entity. It is also a trading method used by traditional finance exchanges (TradFi) to execute order books. Bids and offers are matched on a price-time priority basis. The highest bid and the lowest ask order define the best market for that specific order. Price discovery and liquidity are generated by the continuous interaction between buyers, sellers, and the market makers. 

CEXs usually require using their own custodial accounts, which are under the exchanges’ control. A CEX uses a computing system owned and operated by the exchange. Crypto assets are traded against fiat currency or among themselves. Fiat (such as US dollar or euro)-to-crypto transactions tend to be less liquid than the crypto-to-crypto transactions. Like TradFi exchanges, CEXs offer market participants transparency, liquidity, and real-time market access. However, there are limitations to CEXs due to a reliance on the integrity and fairness of the exchanges.

In DEXs, there is no central third party that oversees and controls the trading. Instead of using an order book system where price discovery relies on matches between buyers and sellers based on price and volume orders, DEXs use an Automated Market Maker (AMM) protocol. AMMs allow permissionless, public, on-chain trading based on smart contracts that set prices depending on the liquidity provided by market participants. The protocol is automated and cannot be censored or altered by a third party. 

Liquidity pools are funded by liquidity providers (LPs) that deposit their tokens for one or more cryptocurrencies. For a two-cryptocurrency pool, the number of tokens in each currency is calculated such that the value of the pool is evenly split between the two. For providing liquidity, LPs are rewarded with a liquidity token or a fee every time a trade is executed. 

A trader may sell a crypto asset in exchange for the other crypto asset and pay a fee. There are varying models/designs to AMMs in DEXs, and one of the foundational ones is the Constant Product Market Maker (CPMM) in Uniswap V2. The function used by CPMM for the price of the trade based on the available liquidity in a pool is the constant product. The product of the number of coins in each of the two cryptocurrencies stays the same before and after the trade (assuming no fees). The price, though, of one currency in terms of the other, changes post-trade.

Additionally, a fee is added to the liquidity pool and will change the constant product to a new value. Due to the changing ratio between the tokens after a trade, if an LP redeemed a portion of their position from the liquidity pool, it may be subject to a loss (referred to as impermanent loss) compared to the alternative of passively keeping their tokens in a wallet. Such price movements caused by trades in an AMM protocol provide opportunities for arbitrageurs who take advantage of the price differentials between different platforms and contribute to price discovery. 

DEXs cover exchanges of spot digital assets where there are two roles: trader and LP. Uniswap V1, launched in 2018, supported swaps only between ETH and ERC-20 (standard for Fungible Tokens). It was built on the Ethereum network. V2 introduced swaps between an ERC-20 with another ERC-20. V3 introduced the concept of concentrated liquidity, where a provider can specify a price range within which to add liquidity. Within a price band, for small enough trades, V3 functions following the CPMM mechanism of V2. For larger trades, the contract is split and needs to be executed in different price bands where new amounts of liquidity are provided. V3 also introduced a flexible fee structure. Notably, any user can create a liquidity pool, and there can be multiple pools with the same pair and fee. All three versions of the Uniswap protocol contracts continue to function.

Below is a worked example to demonstrate some of the features of DEXs.

For any given token amounts, (x,y) the CPMM uses the function x*y=k to manage all trades between the two tokens. Assume we have a liquidity pool with two tokens: X (for example ETH) and Y (an ERC-20 token). Assume that the price of 1 ETH is 100 Y. An LP would typically add to the pool equal dollar amounts of ETH and token Y, for example, 10 ETH tokens and 1,000 Y tokens. Thus, the constant k in this case is 10*1,000=10,000.

If the LP chooses to add liquidity to this pool, they must add equal dollar amounts of ETH and Y. For example, they can add (1 ETH, 100 Y) to the pool, thus changing the constant product of the pool but not the price. The new constant product would now be 11*1,100=12,100.

Assume that the LP does not add liquidity and that we have a pool with (10 ETH 1,000 Y) tokens. A trader wishes to sell 1 ETH in exchange of token Y. Without transaction fees, the pool would now have 11 ETH, and to maintain the invariant product of 10,000 for the number of tokens in the pool, the amount of token Y would decrease to 10,000/11=909.09 tokens (we use a two-decimal rounding). Hence, the trader gets 1000-909.09=90.91 tokens Y.

Assuming a transaction fee of 0.3%, the trader needs to pay 0.003 ETH to execute their sell order. Hence, the number of ETH tokens for the CPMM exchange function is 10.997 ETH (to account for the cost). To preserve the constant product, the amount of Y in the pool would decrease to 10,000/10.997=909.34, and the trader gets 1,000-909.34=90.66 tokens Y. As expected, this is lower than what they would get in the ‘no fee’ scenario. In other words, the trader gets less because they need to pay a transaction fee. In real life, there also is a gas fee, which is paid to Ethereum miners to process the transaction. The higher gas fees one pays, the quicker the trade gets executed. For this example, we will skip gas fees. Notably the trader was expecting for the 1 ETH sold 100 Y tokens and they got only 90.66 tokens Y. Slippage is the percent difference between the expected number of tokens and the actual amount that the trader gets after the trade (100-90.66)/100 = 9.34%.  

Post-trade, the pool has 11 ETH and 909.34 Y, and the new quoted price for Y is 11ETH/909.34Y=0.012ETH/Y while the pre-trade price was 0.01 ETH/Y. As the pool’s supply decreased for token Y, it appreciated in value with respect to ETH. 

The liquidity pool demographics have changed and so has the Constant Product. The pre-trade invariant k=10,000 becomes k’=11*909.34=10,002.73 after the trade, as the pool now has 11 ETH and 909.34 Y. 

If the LP decides to exit the pool after the trade is executed, they will receive 11 ETH tokens and 909.34 Y tokens for a total value of 11 ETH + 909.34 Y*0.012ETH/Y= 22ETH.

Given the appreciation of Y with respect to ETH, if the LP kept their tokens in a wallet and did not participate in the liquidity pool, the total value of the position would be 10 ETH +1,000 Y*0.012ETH/Y= 22.09 ETH.

The difference of 0.09 ETH between the two scenarios is the impermanent loss that the LP realizes, should they choose to redeem their position in the liquidity pool. If the provider does not exit their position, and the price reverts, the loss does not materialize. However, this potential loss could disincentivize market participants from supplying liquidity in AMM protocols. That is why introducing concentrated liquidity mitigates the impermanent loss.

Figure 24 is a simple visualization of the example. 

Figure 25 shows various scenarios ranging from selling one to ten Ether into a hypothetical pool (Initial Pool State = 10 Ether and 1,000 Y token and a Fee of 0.3%)

Figure 26 shows the rapid increase in price of token Y, and of impermanent loss for the LP for various quantities of ETH sold.

Figure 27 shows the number of tokens Y that can be bought for various sell orders of ETH and how expensive this transaction becomes when the quantity of ETH sold increases. The right axis shows the increase in slippage after a trade for various quantities of ETH sold, which is undesirable. These limitations have been addressed in V3 and more recent AMM mechanisms.

Figures 28 through 32 show the volume distribution on various exchanges in descending order for BTC, ETH, DAI, USDC and USDT.