What is a prediction market?

The short answer

A prediction market is a marketplace where people buy and sell contracts tied to the outcome of a future event. Instead of trading shares in a company, you trade a claim on something that either will or won't happen: a candidate wins an election, inflation comes in above 3%, a film takes Best Picture. When the event resolves, the contract pays out a fixed amount if you were right and nothing if you were wrong.

The reason anyone pays attention to these markets is what the price does. Because a correct contract pays a known, fixed amount, its trading price behaves like a probability — a live, money-backed estimate of how likely the event is. When traders think an outcome is nearly certain, they bid its contract up toward the maximum payout. When they think it's a long shot, the price sinks. The market is, in effect, a continuously updating forecast that anyone can read at a glance.

How a contract actually works

The clearest way to understand a prediction market is to follow a single contract. Most markets are binary: they ask a yes-or-no question and settle at $1 for the winning side and $0 for the losing side. A market might ask, "Will the Federal Reserve cut rates at its September meeting?" You can buy YES or you can buy NO.

Say YES is trading at 62¢. You buy one YES contract for 62¢. If the Fed cuts, your contract is worth $1 at resolution and you keep the 38¢ difference. If the Fed holds, your contract expires worthless and you lose the 62¢ you paid. Whoever sold you that contract — the person on the NO side — has the mirror-image outcome. Every dollar one side wins, the other side loses. Nothing is created; the pot is simply redistributed to whoever read the future correctly.

You don't have to hold a contract until the event resolves. Prices move constantly as news arrives, and you can sell at any time for whatever the market will currently pay. Buy YES at 62¢, watch a strong jobs report push it to 78¢, and you can sell for a 16¢ gain without ever waiting to see what the Fed does. In that sense a prediction market feels a lot like a stock exchange — an order book, bids and asks, a price that ticks up and down — except the thing being priced is a fact about the world rather than a company's earnings.

Why the price is a probability

This is the part that makes prediction markets genuinely useful, so it's worth slowing down on. Consider a trader deciding whether YES at 62¢ is a good deal. If they believe the true chance of a rate cut is higher than 62%, buying looks profitable on average: they're paying 62¢ for something that, in their estimation, is worth more. If they think the real chance is lower than 62%, they'd rather sell — or buy NO. Buyers push the price up, sellers push it down, and the price settles where the marginal trader is indifferent. That equilibrium price is the crowd's collective estimate of the probability.

So a contract trading at 62¢ is the market saying, in effect, "there's about a 62% chance this happens." A contract at 4¢ says roughly one chance in twenty-five. A contract at 96¢ says near-certain. You read the odds straight off the price tag — no converting from fractional or American odds, no decoding a point spread. The number in cents is the probability in percent.

There's an elegant consistency check built in. The two sides of a binary market should add up to about a dollar, because exactly one of them will be worth $1 at the end. If YES is 62¢, NO should be around 38¢. When the two sides sum to noticeably more than $1, the extra is the spread — the market's cost of doing business, and a real drag on your returns that's easy to forget when you're focused on being right.

Where the idea comes from

Betting on future events is ancient, but the modern idea — that a market price can be a serious forecasting tool — is more recent. Informal betting markets on elections operated for over a century, and academic and online experiments from the late 1980s onward showed that market prices often predicted outcomes as well as or better than polls. The underlying principle is sometimes called the wisdom of crowds: aggregate a large number of independent, informed guesses and the errors tend to cancel out, leaving a surprisingly accurate signal.

What sharpens a prediction market beyond a simple poll is that participants have money at stake. A poll records what people say; a market records what they're willing to bet. That skin in the game does two things. It punishes idle opinion — being loudly wrong costs you — and it rewards people who actually know something, because they can profit by correcting a mispriced contract. The market pays for information, so information flows in.

Where you'll find them today

Two venues dominate most conversations about prediction markets right now. Polymarket runs on public blockchain infrastructure and settles trades in USDC, a dollar-pegged stablecoin, which lets it list a huge range of markets — politics, crypto, sports, culture, breaking news — often within hours of a question becoming interesting. Kalshi operates as a US-regulated exchange, a designated contract market overseen by the CFTC, settling in ordinary US dollars from a bank account; its catalog leans toward economic and structured recurring questions like inflation readings, interest-rate decisions, and weather.

There are others, and the landscape shifts as regulation evolves, so treat any specific detail here as a starting point to verify rather than gospel. The important thing for understanding the concept is that the mechanics are the same across venues: a yes-or-no question, contracts that pay a fixed amount, and a price that reads as a probability. Learn it on one platform and the others will feel familiar.

What they're good at

Prediction markets shine at producing a single, legible number that updates in real time. A poll is a snapshot that's stale the moment it's published; a market price moves the instant new information lands. During a fast-developing story — an election night, a central-bank announcement, a court ruling — the market often reprices within seconds, giving you a running estimate that no periodic survey can match.

They're also honest in a specific way. Because being wrong costs money, the price tends to strip out wishful thinking and partisan noise. Pundits face no penalty for a confident bad call; a trader does. That doesn't make markets infallible, but it does mean the number reflects conviction weighted by willingness to pay, which is usually a better guide than the loudest voice in the room.

Where they fall short

None of this makes a prediction market a crystal ball, and it helps to know the failure modes before you lean on one:

• Thin markets lie. A contract with only a few thousand dollars of volume reflects a handful of opinions, not a crowd. The price is a quote, not a consensus, and it can be moved by a single motivated trader.
• Long shots are overpriced. Very cheap contracts — a few cents — tend to trade above their true probability, the same way lottery tickets do. People like cheap bets with a big payoff, and that demand inflates the price.
• Resolution is everything. A market is only as good as the exact wording of what counts as the event happening. Ambiguous criteria, a specific time zone, or a source of truth that reports on a lag can all mean the contract settles differently from how a casual reader expected.
• Fees and spreads eat edges. The gap between buy and sell prices, plus any trading or withdrawal fees, is a cost you pay whether or not you're right. A small forecasting edge can vanish entirely once you account for it.
• They can be manipulated or thinly informed. On a low-volume market, someone can push a price to create a misleading headline. And on genuinely uncertain questions, the crowd can simply be wrong together.

How to actually read one

If you're looking at a prediction market for the first time, start with three questions. What exactly does this contract pay out on — what's the precise resolution criterion? How much volume and depth does the market have, so you know whether the price is a real consensus or a thin quote? And do YES and NO add up close to a dollar, or is the spread wide enough to erode any edge you think you have?

From there, the discipline that separates useful analysis from gambling is simple to state and hard to practice: before you take a side, name the probability you disagree with and why. The market says 62%. Do you genuinely believe the real number is meaningfully different — enough to clear the spread and the fees — and can you say what the market is missing? If you can't articulate that, you don't have an edge. You have an opinion, and the market is full of those already.

Used well, a prediction market is less a place to bet and more a place to read: a compact, real-time, money-backed summary of what a lot of motivated people currently believe about the future. That's a genuinely valuable thing to be able to glance at — as long as you remember it's a forecast made of prices, with all the blind spots that implies, and not a promise about what comes next.