Central limit order book

Every venue in this catalogue that calls itself an exchange rather than a game is running this, and the phrase is doing more work than it looks like it is doing. It is not a feature and not a quality claim. It is a statement about who is on the other side of your trade, and everything else on the card follows from it.

How it works

Two queues per outcome. Bids from people willing to pay, asks from people willing to sell, each sorted by price and then by the moment it arrived, and the venue's only job is to pair the best bid with the best ask when the two meet. Polymarket US states the consequence in one sentence of its own documentation: it operates a central limit order book, your order is matched against another user's order, and the platform does not take the other side of your trades and does not set prices.

That last clause is the definition. A venue running this shape holds no inventory, publishes no quote of its own and has no view. Kalshi, Polymarket, Polymarket US and Limitless all run it. ForecastEx runs the nearest variant: bids are ranked by price-time priority, and when a Yes bid and a No bid sum to $1.00 the exchange pairs them, with each side's contract struck against the clearing house rather than against the other bidder.

The number on the screen is usually not a price you can take

Polymarket displays the midpoint of the best bid and the best ask, and falls back to the last traded price once the spread is wider than ten cents. Best bid 34, best ask 40, screen says 37. You will pay 40 or receive 34, and 37 is the one price at which nobody is standing.

In a binary market, one side's bid is the other side's ask

Kalshi's order book endpoint returns bids only, for Yes and for No, because a Yes bid at X is a No ask at a dollar minus X. Read one side of that response, conclude there are no sellers, and you have missed every seller in the book.

A book can be empty, and an empty book has no price

This is the property that separates it from everything else in the liquidity_model enum. A curve always quotes, because a curve is a formula. A parimutuel pool never quotes, because there is nothing to trade until the pool closes. A book quotes when somebody is standing there and falls silent when nobody is, and the silence is honest: the venue is not withholding a price, it is declining to invent one.

What the catalogue records, and the two fields people confuse

economics.liquidity_model: clob is set on 46 of the 60 cards here, amm on 6, none on 8. Read it as a fact about the market behind the card, not about the card's own software: a Polymarket client, a cross-venue feed or a dashboard sets clob because Polymarket runs a book, which is what makes the data arrive as two-sided ladders with a spread and resting size.

capabilities.order_book is a different claim and it is true on 39 cards. It says this tool gives you the book — the ladder, the levels, the resting size — rather than a single price. The two fields come apart, and the cases where they do are the interesting ones: Futuur, Myriad and Manifold all carry order_book: true alongside liquidity_model: amm, because each runs real order types over a curve that catches whatever the queue does not. Five of the six AMM cards in the catalogue are in that position.

Why it matters here

You will read it as a depth claim and size a position on it. It is not one. A book with one resting order at 12 cents is a central limit order book, and so is a book with nothing in it. Where the liquidity comes from has the four numbers that actually settle the question, and the measured gap between two working venues on the same day was about three orders of magnitude.

You will assume you can get out. On a curve you can always close, at a price. On a book you can close only if somebody is bidding, and on a thin market at three in the morning nobody is. Holding to resolution is the default exit on most of this catalogue, not the fallback.

You will trade at the midpoint in your head. Every backtest written against a displayed price rather than against the far side of the spread is optimistic by half the spread on entry and again on exit, which on the Futuur market we measured at a 6-cent spread is more than three times the largest trading fee in the sector.

You will price the fee and forget the book. The taker fee is published and the spread is not, and on a thin market the spread is the larger number. What a trade actually costs converts the published figures into the effective fee at fifty cents that every card carries, and then says plainly that on a thin market it is not the largest part of what you paid.