Most people who’ve traded on a centralized exchange and a DEX in the same week haven’t thought carefully about what’s structurally different between them. The price chart looks similar. The mechanics are not.
On a centralized exchange like Coinbase or Binance, price emerges from a live ledger of posted orders — buyers and sellers declaring what they’ll pay and accept. On a DEX like Uniswap, there are no orders. Price is determined by a formula applied to pool reserves. These aren’t variations on the same idea. They’re two different systems for solving the price-discovery problem, and the difference determines where costs hide, who captures value, and what breaks under stress.
How Order Books Form Prices
An order book is a ledger of open intentions. It holds two sides: bids (buy orders) and asks (sell orders), each with a price and a quantity. When a bid and an ask match on price, the trade executes. When they don’t, the orders sit in the book and wait.
The gap between the best bid (highest buyer price) and the best ask (lowest seller price) is the spread. Tight spreads indicate a liquid market with competing buyers and sellers. Wide spreads indicate thin liquidity or low interest. Market makers — firms that post orders on both sides simultaneously — exist specifically to narrow this spread and capture it as revenue. Every time you cross the spread with a market order, some fraction of that gap flows to the market maker.
Order books are the dominant structure in traditional finance: equities, futures, and foreign exchange all run on variants of the central limit order book (CLOB). On-chain CLOBs exist — dYdX on its own Cosmos-based chain, OpenBook on Solana — but they require high transaction throughput. Every order placement, modification, and cancellation is an on-chain state change. On most EVM chains, this is economically prohibitive. Solana’s architecture makes on-chain CLOBs viable there in a way that Ethereum mainnet does not.
How AMMs Form Prices
Automated market makers eliminate the order concept entirely. Instead of a book of pending trades, two tokens sit in a liquidity pool governed by a mathematical formula. The most common is the constant product formula:
x × y = k
Where x and y are the pool’s token reserves and k remains constant. When you swap token A for token B, you deposit A into the pool and withdraw B. The formula calculates exactly how much B you receive based on how the new reserve ratio satisfies the constant. No counterparty required. No order book. Price is a direct function of reserve ratios.
The “market maker” role is replaced by liquidity providers (LPs) — anyone who deposits equal values of both tokens into the pool. LPs earn a percentage of every swap that flows through. In return, they’re exposed to impermanent loss: when the price of one token moves significantly relative to the other, the pool automatically rebalances by selling the appreciating asset and buying the depreciating one. The LP ends up holding less of what went up and more of what went down, compared to simply holding the tokens outright. The “impermanent” qualifier means this loss reverses if prices return to the deposit ratio — but often they don’t.
AMMs also have price impact: larger trades shift reserves further from the current ratio, worsening the effective price received. On thin pools, even modest trades can move price significantly. This is analogous to market impact on an order book, but the mechanism is automatic rather than driven by order depletion.
Where the Costs Live
On an order book: the cost is the spread (if you’re taking liquidity with a market order), plus any exchange fee. Market makers bear inventory risk and infrastructure costs; traders bear the spread. The cost is visible — limit orders let you set your price and wait for a fill.
On an AMM: the cost is the fee (typically 0.01% to 1% depending on pool tier) plus slippage from price impact. LPs bear impermanent loss; traders bear the fee and slippage. The cost is partially hidden in the execution price — what you see before a trade may differ from what you receive after it, particularly on thin pools or large trades.
Both systems can exhibit MEV (maximal extractable value). On order books, this manifests as front-running. On AMMs, it shows up as sandwich attacks — a bot detects a pending swap, front-runs it to push the price up, then back-runs it to sell at the inflated price. The AMM’s deterministic, public pricing formula makes sandwich attacks structurally easier to execute than on order books with randomized order flow.
What’s Changed Since the Basic Formula
The original Uniswap constant product formula was elegant but capital-inefficient: liquidity was spread uniformly across all possible prices from zero to infinity, most of it sitting at prices the market never visits.
Uniswap v3 introduced concentrated liquidity: LPs specify a price range within which their capital is active. Capital concentrated near the current price is far more efficient — the same notional position captures more fees because it’s all deployed in the active range. The tradeoff is that if price moves outside the specified range, the LP earns no fees and holds a single token (fully converted to the depreciating side).
Curve Finance uses a different formula — the stableswap invariant — optimized for pairs that should trade near 1:1 (USDC/USDT, stETH/ETH). The curve stays flatter at the peg price, allowing large swaps with minimal slippage. It’s not a universal formula but a purpose-built one for correlated assets.
Hybrid designs are emerging: some protocols run AMM pools alongside order books, letting passive LP capital provide baseline liquidity while active market makers post orders on top. dYdX’s migration to a Cosmos app-chain reflects a bet that on-chain CLOBs become viable when you control the execution environment.
What Would Confirm This Direction
Concentrated liquidity becoming the default across major DEXs. On-chain CLOB volume growing to parity with AMM volume on high-throughput chains. Institutional market makers deploying systematically to both AMM positions and on-chain order books. Hybrid liquidity — AMM plus CLOB in the same venue — becoming a standard DEX architecture.
What Would Break or Invalidate It
A critical exploit of concentrated liquidity mechanics draining LP capital and reversing institutional adoption. On-chain CLOBs failing to scale economically even on app-specific chains, driving activity back to off-chain matching. Regulatory classification of AMM liquidity provision as unlicensed market-making, forcing structural changes to LP participation. MEV conditions worsening to the point that retail LPs exit entirely.
Timing Perspective
Now: AMMs are the dominant on-chain trading infrastructure. Concentrated liquidity is the current standard, not the basic constant product. On-chain CLOBs are viable on Solana and dYdX’s chain but not yet dominant.
Next: The capital efficiency gap between AMMs and traditional market-making continues to close as concentrated liquidity tooling matures. Hybrid models are an active design space worth watching.
Later: Whether on-chain CLOBs or evolved AMMs come to dominate institutional on-chain trading is unresolved. It depends on throughput scaling and regulatory clarity around LP participation at scale.
What This Covers and What It Doesn’t
This explanation covers the price-formation mechanism — how orders or formulas determine price, where liquidity provider risk lives, and what drives costs on each system. It does not cover how to trade on either system, which to use for a given trade, or whether LP participation makes sense in any context. Those questions depend on factors outside this scope.
The structural distinction between order books and AMMs is foundational knowledge for anyone working with on-chain finance. Both systems are active and evolving. Neither has displaced the other.
