Skip to main content

This site is for educational purposes only. Nothing here constitutes financial advice.

Topic 78 of 189

AMM Curves Compared

How constant-product (Uniswap V2), concentrated-liquidity (Uniswap V3/V4), stable-swap (Curve), and weighted-pool (Balancer) AMMs differ mathematically — and which to use for which asset pairs.

Beginner
8 min readUpdated July 2026Block Clarity Hub Editorial Team

The basics

Why Multiple AMM Designs Exist

Different asset pairs have different trading characteristics. Two highly correlated stablecoins (USDC/USDT) should trade at ~1:1 with very tight spreads. Two volatile assets (ETH/BTC) need much wider price ranges. A token with a known fixed weight relationship to others needs yet another model. AMM (Automated Market Maker) designs have specialised over time to handle these different cases. Picking the right AMM for an asset pair has substantial impact on capital efficiency and trader experience.

Constant Product (Uniswap V2)

The original Uniswap V2 formula: x × y = k. The pool contains x units of token A and y units of token B; their product k stays constant after every trade. Mathematically elegant; works for any token pair; but capital is spread across the entire price range from 0 to infinity. Only a small portion of capital is 'doing work' near the current price. Gas-efficient and simple to integrate; the foundation of DeFi but increasingly displaced by more specialised designs.

Concentrated Liquidity (Uniswap V3, V4)

Uniswap V3 (May 2021) introduced concentrated liquidity. LPs specify a price range where their capital is active. Inside that range, capital is much more efficient — for stablecoin pairs concentrated near 1:1, V3 LPs can offer effective depth comparable to V2 pools with ~100x the capital. The catch: LP positions become directional bets. If price exits your range, you accumulate the 'losing' asset and stop earning fees. Active management is required for optimal returns.

  • Different asset pairs need different AMM curves
  • Constant product (V2): elegant, universal, but capital-inefficient
  • Concentrated liquidity (V3, V4): efficient within a range, but requires active LP management
  • Stable-swap (Curve): designed for highly correlated assets
  • Weighted (Balancer): supports multi-asset pools with custom weights

Key Takeaways

  • AMM design choice has substantial impact on capital efficiency and trader prices
  • Constant product is universal but inefficient; concentrated liquidity is the modern default
  • Specialised curves (Curve for stables, Balancer for weighted pools) serve specific use cases
  • LP positions in advanced AMMs are directional — managing them is part of the LP job

Going deeper

Curve's StableSwap Formula

Curve's StableSwap is designed for pairs that should trade near 1:1. The curve is much flatter than constant product near 1:1 and only steepens as it approaches the edges. This produces very low slippage for small-to-medium trades within the flat region, at the cost of higher slippage if the peg breaks. The mathematical formula combines constant product (for safety at extremes) with constant sum (for efficiency near peg). The StableSwap invariant is what makes Curve dominant for stablecoin and LST trading.

Curve V2 — Crypto-Crypto Pools

Curve V2 generalises the StableSwap idea to volatile asset pairs. The curve tracks a moving 'price oracle' internally and concentrates liquidity around it. This makes Curve V2 pools (TriCrypto, tricrypto2, etc.) competitive with concentrated liquidity AMMs while keeping LP positions passive. V2 pools dominate trading for ETH/BTC and similar correlated-but-volatile pairs. The math is substantially more complex than V1.

Balancer's Weighted Pools

Balancer extends constant product to multi-asset pools with custom weights. Instead of 50/50 ETH/USDC, you can have 80/20 ETH/USDC, or 33/33/34 ETH/BTC/USDC, etc. The weights determine how much each asset contributes to the pool value. Balancer pools double as portfolio-rebalancing tools — if ETH rises, traders sell to the pool, and the pool's automatic rebalancing maintains target weights. Used for index-fund-like exposure and treasury management as much as for AMM trading.

Uniswap V4 Hooks

Uniswap V4 (2024) added 'hooks' — custom logic that runs at specific points in a swap's lifecycle. Hooks let pool creators implement features like dynamic fees, custom oracle behaviour, time-locked withdrawals, and arbitrary on-chain conditions. This makes V4 a 'platform' rather than just a curve — many AMM innovations from the past few years can be expressed as V4 hooks rather than separate protocols. The long-term impact on AMM diversity is significant.

Key Takeaways

  • Curve StableSwap is dominant for correlated assets — much lower slippage than constant product near peg
  • Curve V2 extends to volatile pairs with internal price oracles and concentrated liquidity
  • Balancer's weighted pools enable multi-asset structures and serve as portfolio tools
  • Uniswap V4 hooks make AMM design programmable — many AMM variants can be V4 hooks

Under the hood

Impermanent Loss Across Designs

Impermanent loss (IL) — the gap between LP returns and just-holding returns — varies across AMM designs. Constant product: IL scales smoothly with price ratio changes. Concentrated liquidity: IL is dramatically higher inside your active range (because you have more capital working) and accumulates only the 'losing' asset if price exits. StableSwap: IL is minimal near 1:1 because the curve is flat there. Balancer weighted: IL is reduced for the higher-weighted asset (less exposure to changes). LP profitability depends on whether trading fees outpace IL — different curves have different fee/IL tradeoffs.

Just-In-Time (JIT) Liquidity

In concentrated liquidity AMMs, sophisticated LPs can deposit liquidity moments before a known incoming trade lands and withdraw immediately after. They capture the trade's fees with minimal IL exposure. This 'just-in-time' liquidity provision is a form of MEV; legitimate liquidity providers face being pushed out by JIT predators. Some protocols (Uniswap V4 via hooks) can mitigate JIT; others accept it as a feature of efficient capital allocation.

Toxic Flow vs Informed Flow

AMM LPs face two types of trades. Uninformed flow: random retail trades that bring no information; LPs profit from spreads. Informed flow: trades by people who know the price is about to move (arbitrageurs reacting to off-chain news, MEV bots exploiting price differences). LPs lose to informed flow because the trade leaves them holding the losing asset. The mix of informed vs uninformed flow determines LP profitability. Stablecoin pairs are mostly uninformed; volatile-asset pairs have substantial informed flow. This shapes whether different AMM types are profitable for LPs.

DEX Aggregators and Multi-Hop Routing

Most volume now flows through aggregators (1inch, Matcha, ParaSwap) and intent-based protocols (CoW Swap, UniswapX) that route across multiple AMMs and liquidity sources. A single user trade might be executed across Uniswap V3 + Curve + Balancer + private market maker, picking the optimal path. This makes 'which AMM is best' less relevant for end-user pricing — aggregators handle the routing. The role of specific AMMs is increasingly as liquidity primitives rather than user-facing trading venues.

LP Profitability Requires Honest Analysis

Many LP yields advertised on protocols are gross fee yields not adjusted for IL. Actual net returns (fees minus IL) are frequently lower or negative, especially for volatile pairs. Before LP'ing into any AMM, calculate the IL you would have incurred over historical price paths and compare to fees earned. Tools like Revert and Apy.vision specifically calculate net LP returns. The gap between 'yield farming yield' and 'actual return' is often dramatic.

Key Takeaways

  • IL varies by AMM design; concentrated liquidity amplifies it, StableSwap minimises it near peg
  • JIT liquidity is a form of MEV that can push out legitimate LPs in concentrated liquidity pools
  • AMM LP profitability depends on mix of informed vs uninformed flow
  • Aggregators route across multiple AMMs; 'best AMM' matters less for end-user pricing than for liquidity provision
  • Always calculate actual LP returns net of IL — advertised yields are gross

Related Content

Delta-Neutral Stablecoins (Ethena sUSDe)

How Ethena's sUSDe captures perpetual funding rates by hedging spot crypto with short perpetuals, what makes it different from algorithmic stablecoins, and the structural risks beneath the surface.

Liquid Staking Tokens and Restaking

How stETH/rETH liquid staking tokens work, why EigenLayer's restaking model created a new economic primitive, and what AVS economics mean for the broader Ethereum ecosystem.

Options Greeks On-Chain

How delta, gamma, vega, and theta translate from traditional options markets to on-chain protocols like Derive (formerly Lyra), Aevo, and Premia — and what's structurally different about decentralised options markets.

TWAP and VWAP Execution Algorithms

How time-weighted and volume-weighted average price execution algorithms split large orders to minimise market impact, and what their crypto-native equivalents look like on DEXs and CEXs.

Volatility Products

How variance swaps, volatility tokens, and structured products let users trade volatility itself rather than direction — and the on-chain protocols building this category.

Dollar-Cost Averaging Mechanics

How DCA actually performs mathematically vs lump-sum investing, what the empirical evidence shows for crypto specifically, and the on-chain protocols that automate it.

Portfolio Rebalancing Systematics

How systematic rebalancing keeps portfolio weights at target allocations, the math of why rebalancing produces an excess return in volatile markets, and on-chain rebalancing protocols.

Prediction Market Mechanics

How Polymarket, Augur, and Kalshi structure event-betting markets, what makes a 'good' prediction market, and the structural questions around accuracy, manipulation, and information aggregation.

References & further reading