Price versus probability

We write "the market prices a cut at 45%" all day. The shorthand is fair: a contract that pays 100 if an event happens and trades at 45 is, in effect, the crowd saying the event is a little less likely than not. But a price is a trade, not a poll, and the difference matters when you read it.

Why a price can be read as a probability

If you believe an event is 60% likely, a YES contract at 45 is worth buying: you expect to receive 100 sixty percent of the time, an average of 60, for a cost of 45. Traders who think that keep buying until the price reaches the level where nobody sees an edge. In a liquid market with many participants, that level is a reasonable summary of what the informed money believes. That is why we quote it.

Four ways a price is not a probability

Thin markets. A contract with a few hundred dollars traded can sit at a stale price for days. The last trade may have been one person's opinion. Look at volume and at the bid-ask spread before reading a probability into it; a market quoting 30 bid and 50 ask does not "say 40%."

Long horizons. Money tied up for two years in a contract paying 100 has a cost. Long-dated markets tend to price unlikely outcomes a little higher than pure probability would justify, because the sellers of those contracts want compensation for locking up capital. A 2028 presidential contract at 3% is not a clean 3% belief.

Fees and rounding. Where fees are charged on winnings, the price traders will pay for a given belief is slightly lower than the belief itself. In the extremes, contracts rarely trade below 1 or above 99 even when the outcome is near-certain, because there is no profit left after costs.

The displayed number. Polymarket's displayed price can be the midpoint of the bid and ask, or the last trade when the spread is wide. Kalshi shows a last-traded price. Neither is necessarily a price you can execute at right now. Our trackers show bid and ask alongside the price for that reason.

A worked example

A market on the Fed holding rates at its next meeting shows 62% with $1.2 million traded and a one-point spread. That is a probability worth quoting: deep, tight, and recent. (The live number is on the Fed decision odds page.)

A market on a specific mathematician winning a specific prize shows 8% with $60,000 traded and a spread of 5 to 12. Quoting "8%" is technically correct and substantively misleading. The honest sentence is "thin trading has it somewhere between 5 and 12."

Whose number is it?

Every probability we print is one of three things, and reads as one: what the market prices, what a named outside model estimates (CME FedWatch, sportsbook consensus, a forecaster), or our own estimate, labeled as such. An unlabeled number should make you suspicious, on our pages or anyone's.

What to take away

• In deep, tight, recent markets, the price is a fair probability read.
• In thin or long-dated markets, read the spread and the volume before the price.
• The displayed price is not always an executable price.
• Ask whose number it is every time.