Decentralized exchanges don't use order books. There's no database of buy and sell orders waiting to be matched against each other. Instead, trades execute against pooled reserves held in a smart contract — and those reserves are deposited by independent participants who earn fees in exchange.
This is the liquidity pool model. It solves a real bootstrapping problem: an order book exchange with no users has no liquidity, and without liquidity, no one will use it. Liquidity pools decouple trading availability from active market-maker participation. The pool is always there; anyone can trade against it at any time.
The tradeoff is that something else has to determine the price. That's where the math comes in.
The Constant Product Mechanism
A liquidity pool holds two tokens in a smart contract. The most common pricing model is the constant product formula: x × y = k, where x is the quantity of token A, y is the quantity of token B, and k is a constant the protocol preserves after every trade.
When a trader swaps token A for token B, they send token A to the contract and remove token B. The formula requires that the product of the reserves stays constant. This means the price of each token is determined entirely by the current ratio of reserves — no external input, no price feed, no order matching required.
Here's what that looks like concretely. Suppose a pool holds 100 ETH and 200,000 USDC. The implied ETH price is 2,000 USDC (200,000 / 100). A trader wants to buy 1 ETH. The new state requires: 99 ETH × new USDC reserve = 20,000,000 (the constant k = 100 × 200,000). New USDC reserve = 20,000,000 / 99 = 202,020.20. USDC required = 202,020.20 − 200,000 = 2,020.20 USDC.
The trader paid 2,020.20 for 1 ETH even though the headline price was 2,000. That 1% premium is price impact — a direct consequence of pool depth. Larger pools have smaller price impact for equivalent trade sizes; smaller pools have steeper slippage.
This is a hard constraint. Price impact is a mathematical property of pool depth, not an implementation choice. The only way to reduce it is more liquidity.
How LP Shares Work
Anyone can deposit liquidity into a pool. When they do, the protocol mints LP tokens — fungible tokens representing their proportional ownership of the pool's reserves.
The deposit must be made in both tokens, at the current pool ratio. If the pool is 50% ETH and 50% USDC by value, you deposit both — half your capital in each. You can't deposit a single asset without shifting the implied price, which would create an immediate arbitrage opportunity exploited within seconds.
LP tokens track ownership of the entire pool, including accumulated trading fees. When you withdraw, the protocol burns your LP tokens and returns your share of whatever the pool currently holds — including all fees earned during your deposit period.
LP token value is calculated as: LP token value = (total pool value) / (total LP tokens in circulation). As trading fees accumulate inside the pool's reserves, the total pool value rises while the number of LP tokens stays constant. Each LP token becomes worth slightly more. This is how fees compound — you don't need to claim them manually; they inflate your position continuously with every trade that passes through the pool.
The Impermanent Loss Problem
Providing liquidity comes with a specific risk that doesn't exist for passive holders: impermanent loss. It's worth understanding precisely, because it's often described vaguely when it has a clear mechanical explanation.
When you deposit into a pool, you fix your exposure to both tokens at the current ratio. The constant product formula then automatically rebalances the pool as prices shift — effectively selling the token that's appreciating and buying the one that's declining. The result: you end up with less of the token that went up and more of the one that went down, compared to what you'd have if you'd simply held both tokens without depositing.
The "impermanent" label reflects the fact that if prices return to their original ratio, the loss disappears. But if prices don't revert — which is common in trending markets — the divergence is realized when you withdraw.
The math is worth knowing: for a standard 50/50 pool, if one token doubles in price relative to the other, impermanent loss is approximately 5.7%. If it quadruples, approximately 20%. Trading fees need to exceed that loss for liquidity provision to be profitable on a risk-adjusted basis.
This creates a practical filter. Constant-product pools work best for pairs where prices don't diverge dramatically — two stablecoins, correlated assets, or pairs where fee income is high enough to offset directional risk. Pools with highly directional assets create structural losses for LPs over time.
Stablecoin Pools and Concentrated Liquidity
Two structural improvements have addressed the efficiency problems in the basic constant product model.
Stablecoin pools (Curve's StableSwap formula) recognize that for assets pegged to the same value — USDC/USDT, for example — almost all trades happen near the 1:1 ratio. The standard constant product formula distributes liquidity evenly across the entire price range, wasting most of it outside the region where trades actually happen. Curve's formula concentrates liquidity around the peg, dramatically reducing slippage for near-par trades without requiring LPs to manage ranges manually.
Concentrated liquidity (Uniswap v3, launched 2021) extends this logic to any pair. Instead of spreading capital from zero to infinity, LPs specify a price range where their capital is active. An LP confident ETH/USDC will trade between 1,800 and 2,200 concentrates all capital in that band — earning fees only when price is in range, but doing so with capital efficiency up to 4,000x compared to the full-range model.
The tradeoff is active management. If price moves outside your range, the position goes idle and earns nothing. Passive LPs using v3 without active range management often underperform simple token holding. The efficiency gains are real; capturing them requires deliberate position maintenance.
What Would Confirm or Break This Model
Confirmation signals: LP fee revenue continuing to exceed impermanent loss across major pools on a rolling basis. Concentrated liquidity tooling — automated range vaults, strategy managers — becoming accessible enough for less active LPs. DEX volume maintaining or growing its share of total spot trading.
Invalidation signals: A smart contract exploit draining a major AMM's reserves (which has happened historically in smaller pools). Regulatory action classifying LP provision as unlicensed market-making in major jurisdictions. A structural shift to intent-based protocols routing the majority of volume off-chain, away from AMM execution entirely — though this would likely reduce pool volume rather than eliminate pools as infrastructure.
Timing
Now: Liquidity pools are the backbone of on-chain trading. Hundreds of billions in total value locked across Ethereum mainnet, L2 networks, and alternative chains. The mechanism is live and functioning at scale.
Next (2026): Automated range management for concentrated liquidity positions is still developing. Whether these tools become accessible enough for passive LPs to participate profitably without active monitoring is an open question.
Later: Intent-based trading architectures may route a growing share of high-value volume away from direct AMM interaction. This wouldn't eliminate liquidity pools but would change their role — from primary execution venues to deep liquidity infrastructure that solvers tap into when required.
Boundary Statement
This explains the mechanism. It doesn't address whether providing liquidity is appropriate for any specific pair, position size, or time horizon. Profitability depends on fee rates, price volatility, time in range, gas costs, and opportunity cost — all of which vary significantly by context.
The pool works as described. Whether participating in it makes economic sense is a separate question.
