A liquidity pool is a smart contract that holds two or more tokens so people can trade against it directly. There is no order book and no counterparty on the other side. A formula, known as an automated market maker (AMM), sets the price from the pool's own reserves. Anyone can deposit tokens and become a liquidity provider (LP). LPs receive a share of the pool and of its trading fees. In return they carry price risk, including impermanent loss, and the risk that the contract itself fails.
How a liquidity pool differs from an order book
On an order-book venue, buy and sell orders wait until a matching order arrives at an agreed price. A pool works the other way round. The academic survey SoK: Decentralized Exchanges (DEX) with Automated Market Maker (AMM) Protocols (Xu et al., ACM Computing Surveys, 2023) calls this a "peer-to-pool" design: traders swap with the pool's reserves, not with another person.
| Order book | Liquidity pool (AMM) | |
|---|---|---|
| Who sets the price | Buyers and sellers placing orders | A formula applied to the pool's reserves |
| Who supplies liquidity | Market makers posting orders | Anyone who deposits tokens into the pool |
| What a trade needs | A matching order on the other side | Enough reserves in the pool |
| What a trade costs | Spread and fees | Pool fee plus price impact, which depends on pool depth |
On a decentralised exchange you never hand your tokens to an operator. They stay in your wallet until you send them to a contract (ethereum.org).
How the pricing formula works
The constant product rule: x · y = k
The best-known AMM design keeps the product of the two reserves constant. In this formula, x is the amount of token A, y is the amount of token B, and k has to stay the same after every trade, before fees. Vitalik Buterin described this "x*y=k market maker" on Ethereum Research on 2 March 2018. He credited the idea to Martin Köppelmann of Gnosis. In the same thread it was later renamed the "constant product market maker." Buterin's post also noted that a fee, for example 0.3%, can make the market maker profitable.
When someone buys token A, the pool's supply of A falls and its supply of B rises. Because k stays fixed, each further unit of A costs more.
Worked example: buying 1 ETH (fees ignored)
A pool holds 100 ETH and 200,000 USDC, so k = 20,000,000 and the implied price is $2,000 per ETH.
| Before | After buying 1 ETH | |
|---|---|---|
| ETH in pool | 100 | 99 |
| USDC in pool | 200,000 | 202,020.2 |
| k | 20,000,000 | 20,000,000 |
| Implied ETH price | $2,000 | about $2,040.6 |
The buyer pays about 2,020 USDC for 1 ETH, roughly 1% above the starting price. That difference is price impact. Larger trades and smaller pools make it larger.
Arbitrage keeps pool prices in line
When a pool's price drifts away from the price elsewhere, arbitrageurs buy the cheap side and sell the dear side until the gap closes. The SoK survey describes this activity as what rebalances pools. LPs pay for this, as covered under loss-versus-rebalancing below.
LP shares and fees
A deposit of both tokens gives you pool shares, usually issued as LP tokens, in proportion to your contribution. To withdraw, you hand the shares back and receive your portion of the reserves at the time you withdraw, fees included (SoK survey). If you own 1% of the shares, you are entitled to 1% of the reserves and earn 1% of the fees while you stay in the pool.
Fees are a percentage of each swap. A widely used early design charged one fixed 0.30% fee. According to that protocol's v3 whitepaper, its successor launched with three fee tiers, 0.05%, 0.30% and 1%, chosen per pool. Lower tiers are typically used for pairs that move closely together, and higher tiers for volatile pairs.
Some protocols add liquidity mining: extra reward tokens, often governance tokens, paid to LPs who stake their LP tokens. These rewards come from the protocol's token supply, not from trading activity. A headline APY made up mostly of such rewards depends on the reward programme continuing and on the reward token keeping its value.
Types of liquidity pools
| Pool design | How it prices | Typical use |
|---|---|---|
| Constant product | x · y = k, usually 50/50 by value | General token pairs |
| Stable swap | Mixes constant-sum and constant-product curves | Assets with the same peg, such as two dollar stablecoins |
| Weighted (constant mean) | Each asset has a weight; the weights sum to 1 | Pools of more than two assets, or uneven splits such as 80/20 |
| Concentrated liquidity | LPs choose the price range their capital covers | Active LPs who want more fees per dollar deposited |
- Stable swap pools. The SoK survey describes this design as "a combination of constant-sum and constant-product." An amplification parameter controls the curve. The pool stays almost flat near the peg, so slippage is low while both assets hold it.
- Weighted pools. These generalise the constant product to a weighted geometric mean, and a 50/50 two-asset pool is the special case. The survey's example uses eight assets. An 80/20 pool behaves more like holding mostly the larger asset.
- Concentrated liquidity. The v3 whitepaper describes LPs bounding their liquidity within an arbitrary price range. Inside that range their capital earns more fees per dollar. Outside it, their position earns no fees and sits entirely in one of the two tokens.
What does a price move cost you as an LP?
Enter how far one token's price moved since you deposited, and any fees you earned, to see how a 50/50 constant product position compares with simply holding the tokens.
| Price ratio r of one token between deposit and withdrawal (2 = it doubled, 0.5 = it halved) | Fees you earned while in the pool, as a share of what the tokens would be worth if held | Pool position compared with holding (negative = worse than holding) |
|---|---|---|
| 2 x | 0 % | -5.7 % |
| 1.5 x | 0 % | -2 % |
| 4 x | 0 % | -20 % |
| 0.5 x | 0 % | -5.7 % |
| 10 x | 0 % | -42.5 % |
- The price ratio and fee figures are your own inputs, not figures from any source.
- The formula holds only for a 50/50 constant product pool; with no fees, a ratio of 1 means no loss, and a fall gives the same loss as the matching rise (0.5x and 2x both give about −5.7%).
- The loss is permanent only if you withdraw while prices are still apart; it reverses if they return to where they were at deposit.
This is arithmetic on your inputs only. It does not cover stable swap, weighted or concentrated-liquidity pools, reward-token emissions, gas costs, smart contract or peg risk, and it cannot predict prices or fee income.
Impermanent loss, with numbers
Impermanent loss, also called divergence loss, is the gap between the value of your pool position and the value of the same tokens had you simply held them. The SoK survey defines it as L = V′ / V_held − 1. It happens because the formula sells the asset that is rising and buys the one that is falling as prices move.
Example: ETH doubles after you deposit
You deposit 1 ETH and 2,000 USDC when ETH is at $2,000. ETH then rises to $4,000. Fees are ignored here; they offset part of the loss.
| ETH | USDC | Total value | |
|---|---|---|---|
| Held in wallet | 1.000 | 2,000 | $6,000 |
| In a 50/50 constant product pool | 0.707 | 2,828 | $5,657 |
| Difference | −$343 (−5.7%) |
For a 50/50 constant product pool, the loss depends only on the price ratio r between deposit and withdrawal: IL = 2√r / (1 + r) − 1.
| Price change of one asset | Impermanent loss |
|---|---|
| 1.25x | about −0.6% |
| 1.5x | about −2.0% |
| 2x | about −5.7% |
| 4x | −20.0% |
| 10x | about −42.5% |
The loss reverses if prices return to where they were at deposit. It becomes permanent when you withdraw while the prices are still apart. The same figures apply to falls: a 0.5x move gives the same −5.7% as a 2x move.
Do fees cover it? The evidence
Two research results are worth knowing before you supply liquidity.
- Fees vs. impermanent loss in concentrated pools. Loesch, Hindman, Richardson and Welch (arXiv, November 2021) studied 17 concentrated-liquidity pools that together held 43% of TVL on that design. From launch to their cut-off date, LPs earned $199.3m in fees and suffered $260.1m in impermanent loss. They would have been $60.8m better off simply holding the tokens. The authors also found that concentrating liquidity raises both fee income and impermanent-loss risk.
- Loss-versus-rebalancing (LVR). Milionis, Moallemi, Roughgarden and Zhang (Automated Market Making and Loss-Versus-Rebalancing, first posted August 2022, revised August 2026) split LP returns into two parts. One is market exposure. The other is fees minus what arbitrageurs extract from the pool. In their study of a large ETH-USDC constant product pool, more than 99.991% of the variance in LP returns came from market exposure. In practice, an LP position is mainly a bet on the two assets' prices, with a fee stream that has to beat the arbitrage cost.
Other risks
- Smart contract risk. A bug in the pool contract, or in any contract stacked on top of it such as a farm or a vault, can put deposits at risk. Audits reduce this risk but do not remove it.
- Malicious pools and rug pulls. The SoK survey lists rug pulls among application-layer attacks. A pool's creators can drain liquidity they control, or a token can be built so holders cannot sell it.
- Sandwich attacks. A trade has to allow some slippage, and the survey notes that this tolerance "can be exploited to perform e.g. sandwich attacks." An attacker trades just before and just after your swap. Setting a tight slippage limit reduces the exposure.
- Peg risk in stable pools. Low slippage depends on both assets holding their peg. If one loses it, the pool fills up with the weaker asset.
Liquidity pools beyond swaps
The word "pool" is also used for lending. On Aave, suppliers deposit a single asset into a pool and receive interest-bearing aTokens. Borrowers post more collateral than they borrow. Interest rates rise with utilisation, the share of the pool currently lent out, and climb faster above an optimal level (Aave v3 documentation). This kind of pool has no impermanent loss, because it holds only one asset. Its main risks are borrowers defaulting, liquidations failing, and failures in the contracts. Ethereum.org notes that flash loans also draw on funds held in such pools.
How to assess a pool before depositing
- Where does the yield come from? Separate fee income from reward-token emissions. Pool trackers such as DeFiLlama's yields page show both.
- Volume relative to TVL. Fees scale with trading volume. A large pool with little trading earns little per dollar deposited.
- How volatile is the pair? Use the impermanent-loss table to estimate what a plausible price move would cost, and compare it with expected fees.
- Contract history. Check that the code is open source, has a public audit, and has run for a long time without incidents. For concentrated liquidity, also check that you can monitor and rebalance your range.
- Network costs. Every deposit, claim, rebalance and withdrawal is a transaction that costs gas. Frequent activity can eat into a small position.
The pricing curves get a fuller treatment in our explainer on automated market makers, and the loss formula is worked through step by step in what impermanent loss is. If a single-asset position suits you better, start with how staking works.
FAQ
What is a liquidity pool in simple terms?
It is a smart contract holding a reserve of tokens that people can swap against. A formula sets the price from the reserves. People who deposit tokens earn a share of the trading fees.
Can you lose money in a liquidity pool?
Yes. Impermanent loss can exceed the fees you earn; in one 2021 study of 17 pools, it did in aggregate. Contract bugs, malicious pools and stablecoins losing their peg can also cause losses.
Is providing liquidity the same as staking?
No. Providing liquidity to a swap pool usually means depositing two assets and taking on impermanent loss. Staking usually means locking one asset to help secure a proof-of-stake network or to earn protocol rewards. Single-asset staking has no impermanent loss.
