Polymarket is an on-chain prediction market where you trade shares in the outcome of real-world events — and the price of each share is the market’s live estimate of how likely that outcome is. Buy a share for $0.63, and if the event happens you redeem it for $1.00; if it doesn’t, it’s worth $0.00.
That’s the whole idea in one sentence. But how it works under the hood — the order book, the tokens, the settlement — is what makes it interesting to build on, and what creates the data edge we care about. Let’s walk through it.
The core mechanic: shares in an outcome
Every Polymarket market is a question with defined outcomes, most often binary — YES or NO. Each outcome is a tradable share:
- A YES share pays out $1.00 if the event happens, $0.00 if it doesn’t.
- A NO share does the opposite.
Because a share is worth either $1 or $0 at resolution, its price today — somewhere between the two — is simply the market’s implied probability. A YES share trading at $0.63 means the crowd is pricing a 63% chance of that outcome. As new information arrives, traders buy and sell, and the price moves the way a probability does.
You make money the same way you do anywhere: buy a share for less than it turns out to be worth. Think an event is more likely than $0.40 implies? Buy YES at $0.40; if it resolves true, each share redeems for $1.00.
How trades actually get matched
Polymarket uses a central limit order book (CLOB) — the same model as a traditional exchange. Buyers post bids, sellers post asks, and the book matches them at an agreed price.
Here’s the important architectural detail: matching happens off-chain, but settlement happens on-chain.
- Off-chain order book — orders are posted, cancelled, and matched off-chain, so the trading experience is fast and gas-free to quote.
- On-chain settlement — once two orders match, the trade is settled on Polygon (an Ethereum layer-2) in USDC. This is the moment the trade becomes real and irreversible.
This hybrid design gives you exchange-grade speed and on-chain, self-custodial settlement. It also has a consequence that matters enormously for anyone building automated tools — which we’ll get to.
The tech behind the shares
The outcome shares themselves aren’t just database rows — they’re tokens.
- Collateral: USDC. Every market is collateralized in USDC. The system holds real dollars against the shares that can be redeemed.
- Conditional Tokens Framework. Outcome shares are ERC-1155 tokens minted via the Conditional Tokens Framework. Deposit $1 of USDC and you can split it into one YES + one NO share; hold both and you can merge them back into $1. That’s why a YES and NO share in the same market tend to price to roughly $1.00 combined.
- Resolution via an optimistic oracle. When the event concludes, the result is reported and confirmed through UMA’s optimistic oracle — a mechanism where a proposed outcome stands unless someone disputes it with a bond. Once finalized, winning shares redeem for $1.00 each.
If you’ve read our explainer on what the Polymarket mempool is, the split / merge / redeem events above are exactly the non-trade event types you’ll see (and usually skip) in a raw feed.
Why “on-chain settlement” is the interesting part
Here’s the detail that turns Polymarket from a place to bet into a place to build: because every trade settles on Polygon, every trade is a blockchain transaction — and every transaction spends a few seconds in the mempool (the pool of pending, unconfirmed transactions) before it’s written into a block.
That means a Polymarket trade is visible as a pending intention ~3 seconds before it confirms on-chain. Most data sources only report trades after confirmation, so they’re structurally a few seconds behind the market. If you’re building anything reactive — a bot, a copy-trading system, a live signal — those seconds are the entire edge.
Polymarket’s on-chain settlement isn’t just a decentralization feature. It’s the reason a real-time data feed can exist at all.
Where to go from here
Now that you know what Polymarket is and how it settles, the natural next question is how to use that data:
- The conceptual why: What is the Polymarket mempool, and why does it matter?
- The practical how: Stream live Polymarket trades in Python — a five-line client reading every trade in real time.
- Manual or automated? Manual vs. automated Polymarket trading — which approach fits, and where a data edge actually pays off.
Polyflux streams every Polymarket trade from the mempool over a single WebSocket, the millisecond it happens. Grab a key and start reading the live feed — you’ll see the market a few seconds before everyone waiting on confirmations.