What is a prediction market order book?
A prediction market order book is the list of everyone waiting to buy or sell shares in an outcome, arranged by price. It looks like a stock exchange's book, but the thing being traded is a claim that pays $1 if an event happens and $0 if it does not, so every price in the book doubles as a probability estimate. This guide walks through the mechanics from zero: what a share is, what bids and asks mean, how to read spread and depth, and how Polymarket's order book in particular works.
Shares that pay $1 or nothing
Every binary prediction market has two outcome shares: YES and NO. When the market resolves, one of them is worth exactly $1 and the other exactly $0. Before resolution they trade at prices between 0 and 1. If YES trades at 0.62, buyers are paying 62 cents for a claim on a dollar, which only makes sense if they think the event is at least about 62% likely. That is the core trick of prediction markets: the price is the crowd's probability, discovered by trading rather than polling.
The two shares are complementary. Holding one YES and one NO share guarantees exactly $1 at resolution no matter what happens, so YES and NO prices must sum to about $1. If they drifted apart, arbitrageurs could buy both sides for less than a dollar (or mint a pair for $1 and sell both sides for more) and lock in the difference. In practice this keeps NO priced at roughly 1 minus the YES price, and it means buying NO and selling YES are the same economic bet expressed two ways.
Bids and asks as probability quotes
An order book is two sorted lists. Bids are offers to buy, sorted from highest price down; asks are offers to sell, sorted from lowest price up. The highest bid and the lowest ask are the "top of book", and the gap between them is the spread. In a prediction market these quotes read naturally as probability bounds: a best bid of 0.62 and best ask of 0.63 says the market disagrees about the outcome's probability by only one percentage point.
The midpoint, halfway between best bid and best ask, is the usual single-number summary of where the market stands. The spread is the cost of impatience: crossing it (buying at the ask and later selling at the bid) loses you the spread even if the probability never moves. Tight spreads mean active market makers and cheap trading; wide spreads mean uncertainty, low interest, or both.
How Polymarket's CLOB works
Polymarket runs a central limit order book, or CLOB: a single shared book per outcome where anyone can rest limit orders and anyone can trade against them. Orders match by price and time priority, the same rules as a traditional exchange. What makes it a prediction market CLOB is the complementarity described above: an order to buy NO at 0.38 can match against an order to buy YES at 0.62, because together the two buyers fully fund a $1 share pair. The book you see for YES and the book for NO are mirror images of one shared pool of liquidity.
Settlement is the other difference from a stock exchange. When the event resolves, an oracle reports the outcome and winning shares redeem for $1 each. Between open and resolution, the book is the market: every trade, every quote, and every change in sentiment passes through it.
Reading an L2 book: a worked example
"L2" (level 2) means you see every price level with its total resting size, not just the best quote. Here is a small illustrative book, laid out the way an L2 feed presents it, with prices like those you would see on a market from the BTC up or down 5 minute series. Asks sit above, bids below, best prices nearest the middle:
| Side | Price | Size (shares) |
|---|---|---|
| ask | 0.64 | 2,100 |
| ask | 0.63 | 980 |
| spread 0.01, midpoint 0.625 | ||
| bid | 0.62 | 1,250 |
| bid | 0.61 | 3,400 |
Read it like this. The best bid is 0.62 with 1,250 shares resting, and the best ask is 0.63 with 980 shares. The spread is one cent and the midpoint 0.625, so the market's working probability is about 62.5%. If you market-buy 500 shares you get them all at 0.63, since 980 rest there. If you market-buy 2,000 shares you exhaust the 980 at 0.63 and pay 0.64 for the remaining 1,020: your average price slips above the quote. That slippage is why depth matters as much as price.
Notice also the shape. There are more shares bid at 0.61 (3,400) than at 0.62, a common pattern where patient buyers sit one tick behind the front of the queue. Imbalances between total bid and ask depth are one of the most studied order book signals, in prediction markets as in equities. The same structure appears in every market type, from crypto to the per game books in the MLB series, though depth and spread vary widely between them.
What the book tells you that price alone cannot
A price chart compresses the market to one line. The book shows the negotiation behind it: how much conviction sits at each price, how expensive it is to trade in size, and how liquidity appears and vanishes around news. Two markets can both trade at 0.62 while one has $50 of depth and the other $50,000, and they behave completely differently. That is why researchers work with order book data rather than price series when they study market microstructure, and why backtests that ignore the book tend to overstate profits.
Books also change fast. On an exchange the state you see is already stale by the time you read it, and historical analysis requires someone to have recorded every change as it happened. That is what Marketlens does: every individual book change with a millisecond timestamp, anchored by periodic full snapshots, rather than photographs of the book at fixed intervals.
Inspecting a real book in Python
To make this concrete, here is how a real Polymarket book looks through the Marketlens SDK, which serves both the current book and any past state of it. The fields map one to one onto the concepts above; the full response format is on the order book docs page.
from marketlens import MarketLens
client = MarketLens()
book = client.orderbook.get(market_id, depth=5)
print(book.best_bid, book.best_ask)
print(book.spread, book.midpoint)
print(book.bid_depth, book.ask_depth)From here, a good way to build intuition is to pull books for a few markets you know from the data catalog and watch how spread and depth differ between a busy crypto market and a quiet long-dated one.
FAQ
Common questions
Are prediction market prices probabilities?
Close to it. A YES share pays $1 if the outcome happens and $0 if it does not, so a price of 0.62 means the market is willing to trade at odds implying roughly a 62% chance. Fees, spread, and inventory pressure make the price an approximation of the crowd's probability rather than a pure one.
What is the spread in a prediction market?
The gap between the best ask (lowest price anyone will sell at) and the best bid (highest price anyone will pay). A book quoted 0.62 bid, 0.63 ask has a 0.01 spread, meaning one cent per share. Tight spreads signal an actively made market; wide spreads make round trips expensive.
How are YES and NO shares related on Polymarket?
One YES share plus one NO share always pays exactly $1 at resolution, so their prices must sum to about $1. Buying NO at 0.38 is economically the same bet as selling YES at 0.62. Polymarket's order book matches these complementary orders against each other automatically.
What does L2 order book data mean?
Level 2 data shows every price level on both sides of the book with the total size resting at each level. Level 1 is just the best bid and ask. L2 is what you need to judge depth: how much can actually be bought or sold before the price moves.
Try it
Explore real order books
The free tier includes 5M events per day with full API and full archive access, no card required.
$ pip install marketlens