Buy a YES token at $0.65 and you're saying you think there's a better than 65% chance the thing happens. That's the whole premise of a binary market. Two tokens, one for YES and one for NO, each resolving to $1 or $0 when the event concludes. The price of YES at any moment is just the crowd's running estimate of the outcome, priced in real money.
How orders become prices
Most of these markets run a continuous double auction, same as a stock exchange. Buyers post bids, sellers post asks, and when the two cross, a trade prints. The number the platform shows you is usually the midpoint between the best bid and the best ask.
Some platforms use an automated market maker instead. An AMM holds reserves of both YES and NO tokens and moves the price along a bonding curve. Buy a big chunk of YES and the curve pushes the price up to reflect the demand you just added. The common design is a logarithmic market scoring rule, which guarantees you can always trade at some price but needs a subsidy to keep running.
The real difference between the two is slippage. On an order book a large order chews through several price levels and you feel it. On an AMM the slippage is smooth and you can predict it from the curve. For retail-sized bets AMMs usually fill you better. Once your size gets serious, a deep order book wins.
Arbitrageurs are the reason prices stay honest
Prices track reality because arbitrageurs make money forcing them to. If YES is $0.60 on one platform and $0.65 on another, someone buys the cheap one and sells the rich one until the gap disappears. It's constant and it's competitive.
There's also an arb baked right into the contract. YES and NO have to sum to $1 at resolution. So if YES trades at $0.60 and NO trades at $0.35, you buy both for $0.95, collect $1 at settlement, and pocket a risk-free nickel. That's the mechanism that keeps YES plus NO pinned near a dollar.
When you see YES plus NO adding up to more than $1, that overhang is the market maker's fee, the platform's vig. On Polymarket it's often 1 to 2 cents. On thinner platforms it can be 5 to 10 cents, which quietly tells you how much edge you need just to break even before you've been right about anything.
Why limit orders matter more here than in stocks
The bid-ask spread in a lot of these markets runs 3 to 5% of the contract value. A YES token showing $0.50 might have a bid at $0.48 and an ask at $0.52. Hit the market order and you've handed over 2 to 4% of your upside before the event has even played out.
The traders who do this seriously live on limit orders. They post a bid a hair above the current best bid and they wait. In markets where fresh information shows up in bursts rather than a steady stream, that patience pays you back in better average entry prices.
The exception is fast news. When a headline drops and a contract is repricing by the second, resting limit orders get stranded behind the move. In those moments the speed of a market order is worth the cost of crossing the spread, and it's not close.
Settlement is where it gets contentious
Resolution is the interesting part, and sometimes the ugly part. Every contract has predefined criteria, usually pointing at a specific source. A bill-passes-Congress market might reference the official congressional record. A GDP contract might reference the Bureau of Economic Analysis first release.
How it resolves depends on the platform. Centralized venues like Kalshi run internal resolution teams that check the outcome against the stated criteria. Decentralized ones like Polymarket lean on oracles, often UMA's optimistic oracle, where an outcome is proposed and anyone can dispute it inside a challenge window.
Disputes do happen. When the criteria are fuzzy, or the real world coughs up something nobody wrote a rule for, the resolution process gets stress-tested. That's exactly why reading the resolution terms isn't optional. A contract that looks like free money can go to zero if the named source reports it differently than you assumed. I've watched people get this wrong on markets they were technically right about.
The price is also information
The most useful thing about a binary price is what it tells you in aggregate. When thousands of people with real money on the line converge on a number, that number holds information no single person has. Across a wide range of event types, market prices have held up well against expert panels, polling averages, and statistical models.
The mechanism is plain enough. The market rewards being right and punishes overconfidence with your own cash. Think you know better than the price, put money on it. Right and you profit, wrong and you pay for the lesson. Run that loop across thousands of participants and you get prices that are genuinely hard to beat over time.
None of this makes the markets easy to trade. It just means the edge, when there is one, tends to come from doing the boring work: reading the resolution source before you size in, using limit orders where the spread is wide, and knowing which platform is actually deep enough to fill you.