Order Books and AMM Curves: Where Each One Prices Better

In an order book, participants post the prices at which they will buy or sell, and a trade happens when two orders meet. In an automated market maker, a formula sets the price from the ratio of assets in a pool, and anyone can trade against it at whatever that formula says. The first requires someone on the other side to be waiting; the second guarantees a counterparty and lets the price move to compensate.

What the book gives you

Precision. A limit order executes at your price or not at all, which is a control an AMM cannot offer natively — you are choosing certainty of price over certainty of execution.

Depth that can be inspected. You can see the size available at each level and calculate what a given trade would cost before placing it. Whether that displayed depth survives contact with a real order is another matter, but the information exists.

And efficiency at scale. A liquid book concentrates offers near the current price without requiring anyone to lock capital across a range they think is irrelevant, which is why professional market making has always preferred them.

The cost is that the book only works when someone is making it. In a thin market the spread widens until the quoted price is close to fiction, and in a fast one the displayed depth vanishes as makers pull quotes — precisely when you wanted it.

What the curve gives you

A counterparty, always. There is no state in which an AMM refuses to trade; the price simply moves further as the trade grows. For long-tail assets nobody is willing to quote continuously, that guarantee is the difference between a market and no market.

Passive participation. Providing liquidity requires depositing assets and nothing else — no quoting, no infrastructure, no monitoring. That opened market making to people who could never have run a book, which is why AMMs bootstrapped liquidity for thousands of assets that would otherwise have had none.

Transparency of the pricing rule. The formula is public and the reserves are on-chain, so anyone can compute the exact price for any trade size in advance. No book offers that, because a book's future depth depends on the intentions of participants.

The cost is that the curve prices without knowing anything. It has no view on value; it responds only to the trades it receives, which means it is systematically on the wrong side when prices move on external information. Liquidity providers absorb that as a loss relative to simply holding the assets, and it is the fundamental economics of the design rather than a flaw in it.

The size question

Trade size is what usually decides between them.

Small trades in an active market are frequently cheaper on a curve. There is no waiting, price impact is negligible relative to the spread you would cross in a book, and the execution is deterministic.

Large trades are usually cheaper in a deep book, because depth can be concentrated exactly where the trading is happening rather than spread across a range. A curve holds liquidity at prices nobody is trading at, which is capital doing nothing while still counting toward the pool.

Concentrated liquidity narrows this gap considerably by letting providers deploy within a chosen band, which makes a curve behave much more like a book near the active price — and reintroduces the active management that passive provision was meant to avoid.

How they fail

A book fails by emptying. Makers withdraw, the spread widens, and a market that looked liquid a moment ago cannot absorb an ordinary order. The failure is discontinuous and it happens exactly when volatility spikes.

A curve fails by pricing. It never refuses, so a large trade against a thin pool executes at a terrible price rather than not executing. The failure is continuous and visible in advance, which is a meaningful difference: you can compute the bad outcome before accepting it.

Which failure is preferable depends on whether you would rather be unable to trade or able to trade badly. For a liquidation engine that must execute, the curve's guarantee is worth a great deal. For a discretionary trade, the book's refusal is information.

Choosing for a given trade

• Compute the curve's price for your exact size before comparing — it is deterministic and takes seconds.
• For a book, treat displayed depth as an upper bound rather than a promise, especially in volatile conditions.
• Small trades in liquid pairs: the curve is usually cheaper once the spread is counted.
• Large trades: compare against concentrated-liquidity pools specifically, not against a general-purpose curve.
• For long-tail assets there is frequently no choice, and the relevant question becomes how thin the pool is rather than which venue.

The useful framing is that a book prices by asking people and a curve prices by rule. Rules are predictable and uninformed; people are informed and unreliable. Every trade-off between the two designs descends from that.