What Is a Liquidity Pool? How AMMs Work Under the Hood (DeFi 2026)

Why DEXs use pools instead of order books

Traditional exchanges (Coinbase, Kraken, Binance) use order books — sellers post offers at various prices, buyers post bids, the exchange matches them. This works well for high-liquidity markets but requires a centralized intermediary running the matching engine.

On-chain, running an order book is computationally expensive. Every order placement and cancellation is a transaction with gas cost. For low-volume tokens, the market would dry up — no one wants to post limit orders for tiny altcoins. The 2018 invention of Automated Market Makers (AMMs) solved this by replacing the order book with a smart contract holding both tokens. Now anyone can swap by interacting with the contract — instant liquidity for any pair as long as someone has deposited both sides.

How constant product x · y = k actually works

In a Uniswap V2 ETH/USDC pool, suppose the pool starts with 100 ETH (x) and 200,000 USDC (y). The invariant: x · y = k = 20,000,000. Pool price = y / x = $2,000 per ETH.

Now you want to swap 1 ETH for USDC. The contract enforces x · y = k. After your swap: new_x = 101, so new_y = 20,000,000 / 101 = 198,019.8. You receive 200,000 - 198,019.8 = 1,980.2 USDC. Your effective price: 1,980.2 USDC per ETH — slightly worse than the pre-trade $2,000 due to slippage.

The bigger your trade relative to pool size, the worse your effective price. Try the interactive Uniswap V2 simulator and slide the swap amount from 0.5 ETH to 200 ETH to see the bonding curve move dramatically.

How liquidity providers earn fees

When you deposit tokens into a pool, you receive LP tokens — proof of your share. Every swap charges a fee (0.3% on Uniswap V2, 0.05-1% on V3 by tier) that's added to the pool reserves. Your LP tokens are now worth more.

Real-world example: in the USDC/ETH 0.05% pool on Uniswap V3, daily volume of $200M generates $100k in fees. If you're 1% of the pool's liquidity, you earn $1,000/day in fees. Annualized that's 365% — far higher than any bank yield. But there's a catch: impermanent loss. When the price moves significantly, your LP value can fall below what you'd have made just holding the two tokens — a hidden cost we cover in our impermanent loss article.

Uniswap V2 vs V3 — what changed?

V2's constant product spreads liquidity across infinite price range (from $0 to infinity per ETH). In practice, most trading happens in a narrow band — V2 wastes most capital. V3 introduced 'concentrated liquidity': LPs pick a price range [Pa, Pb] where their tokens are active. Inside the range, your capital does 100-4000× more work than V2.

Trade-off: if price moves outside your range, your position is 100% in the worse-performing asset and earns zero fees until price returns. V3 LPs need active management — narrow ranges + bots, or wider ranges + less efficiency. Try the V3 simulator to drag a position range and watch the histogram of liquidity distribution.

What are LP tokens and how do they work?

When you provide liquidity, the contract mints you LP tokens (in V2 these are ERC-20 tokens called 'Uniswap V2 LP'). They represent your share of the pool. To withdraw your liquidity, you burn (return) the LP tokens, and the contract gives you back your share of the current reserves.

LP tokens are themselves valuable. You can stake them in 'farms' (Sushiswap, Convex) to earn additional rewards on top of the swap fees. This is the foundation of 'yield farming'. In V3, LP positions are NFTs instead of fungible tokens — each LP position has a unique range, so they can't all be interchangeable.

Risks of providing liquidity

Three main categories:

• Impermanent loss — when prices diverge, your LP value lags behind HODL. At 2× price move IL is 5.7%; at 4× it's 20%; at 10× it's 42%.
• Smart contract risk — bugs in the pool contract or governance vulnerabilities can drain funds. Pick established protocols (Uniswap, Curve) with audits + long uptime.
• Rug pull risk — for pools created by anonymous teams (especially newly-launched memecoins), the team can drain the pool by removing all liquidity. Only LP into pools with locked liquidity.

Best practices for liquidity providers

1. Start with stable-stable pools (USDC/USDT) — IL is near zero, yields are 3-10% APY. Boring but profitable.
2. Move to ETH/stETH or BTC/WBTC type correlated pairs — minimal IL, slightly higher yield.
3. Only LP into pools with $5M+ TVL on established protocols. Smaller pools = higher IL volatility + higher rug risk.
4. Use V2 (full range) until you understand V3 trade-offs. V3 narrow ranges require active management or you'll lose to V2 LPs.
5. Track your position weekly using DeBank or Zerion. Most LPs lose money because they forget about their position and let bad market conditions drain it.

Uniswap V4 hooks — the next evolution

Uniswap V4 (launched 2024) keeps the V3 concentrated liquidity model but adds 'hooks' — custom smart contract logic that runs before/after pool actions. Examples: dynamic fees that respond to volatility, MEV protection that batches swaps, custom curves that aren't constant-product.

For most users, V4 looks similar to V3 from the trading side. The change is for builders — V4 makes it dramatically easier to launch specialized pools. Try the V4 hooks simulator to see how toggling hooks changes a swap's behavior.