What Is a Prediction Market? How They Work

Q: How is the probability calculated?

It isn't calculated by anyone; it emerges from trading. Because a contract pays $1 if correct, its price in cents is the implied probability. A contract at 72¢ means the market prices the outcome at about 72%.

Q: Are prediction markets accurate?

Often, especially in deep, high-volume markets, where they have matched or beaten polls on some major events. But accuracy depends on liquidity — a thin market’s price can be unreliable or even manipulated.

A prediction market lets people buy and sell contracts tied to the outcome of a future event, such as an election, a Federal Reserve decision, a playoff game, or the date a hurricane makes landfall. Most of these contracts are built to pay $1 if the event happens and $0 if it doesn't. That design has a neat consequence: the contract's price reads directly as the market's collective, money-weighted estimate of how likely the outcome is. A contract trading at 64¢ is the market saying it sees about a 64% chance.

In that sense it behaves like any other market. A stock's price isn't one analyst's call. It's the running, money-weighted consensus of everyone willing to back a view, and it shifts as the view does. A prediction market does the same job with one difference: because each contract settles at either a dollar or nothing, its price reads not as a valuation but as a probability — the crowd's live estimate of whether the event happens.

That single property, price as probability, is what makes prediction markets interesting to anyone whose job is to forecast or to report. They are not a poll, and they are not a model. They are a continuously updating number that real participants are putting real money behind. (For the one-line version, see our glossary entry on prediction markets; this piece goes deeper.)

How a prediction market actually works

Start with the basic unit, the binary, Yes/No contract. Take the question "Will the Fed cut rates at its next meeting?" The market expresses it as two sides, Yes and No, each of which settles at $1 if it's correct and $0 if it isn't. Buy Yes at 30¢, and you stand to make 70¢ if the cut happens. If it doesn't, you lose your 30¢.

Because the payout is fixed at a dollar, the price carries information on its own. In plain terms, the price is the forecast. This is the idea behind implied probability: a Yes contract at 30¢ implies the market thinks there's roughly a 30% chance. As new information arrives, whether a jobs report, a debate, or an injury, participants trade on it, and the last traded price moves. The forecast updates in seconds, without anyone running a fresh survey.

Two other numbers tell you how seriously to take that price.

Trading volume is how many contracts have changed hands over a period. It measures attention and money, meaning how much interest a question is drawing.
Open interest is how many contracts are currently held open, that is, positions entered but not yet closed or settled. Volume is activity over time. Open interest is money at stake right now.

These matter because a price is only as trustworthy as the market behind it. Here's the intuition: a packed, busy market is hard to fool, while a quiet one can be swayed by a single trade. A question with deep liquidity, with lots of resting orders and a tight gap between buyers and sellers, produces a price that's hard to push around and reflects genuine consensus. A thin market can swing on one order. So when you read a probability off a prediction market, always glance at the volume next to it. (For the practical version, see how to read prediction market odds.)

When the event is decided, the market settles. The winning side pays out $1, the losing side $0, and the market closes. Not every market is a simple Yes/No question. Some are mutually exclusive sets, like "who wins the election" or "which team takes the title," where each candidate is its own contract and the prices across all of them sum to about 100%. That gives you a full probability distribution in one view. Platforms structure these as multi-outcome markets.

Why they matter for forecasting and journalism

The case for prediction markets is that they pool scattered information and attach a cost to being wrong. In theory, and often in practice, that produces a sharper, faster-moving estimate than a survey taken last Tuesday.

For reporters and analysts, the appeal is concrete. A market gives you a single, timestamped, quantified probability you can cite and watch move. It reacts to breaking news before the next poll is even in the field. That's why newsroom and research interest in prediction markets has grown alongside the more familiar polling averages and statistical models. They're a complementary signal, not a replacement.

The honest framing is that markets, polls, and models each capture something the others miss. Markets are fast and money-weighted. Polls sample opinion directly. Models impose structure and historical priors. The serious forecaster reads all three. We go deeper on that in prediction markets vs. polls.

A poll asks people what they think for free. A market asks people to back what they think with money, and it weights the loudest voices by how much they're willing to risk.

The major platforms

Three venues anchor the current landscape, and the differences between them matter.

Kalshi is a CFTC-regulated U.S. exchange covering economics, politics, weather, sports, and current events. As a regulated venue, it operates under U.S. market oversight. It's TickerTracker's primary data source, and it carries a live, real-time trade stream.
Polymarket is a large blockchain-based market built on the Polygon network, with positions denominated in the USDC stablecoin. It's known for very high-volume markets on elections and major news. Many are structured as multi-outcome "negative risk" markets, which keep all outcome prices summing to about a dollar.
Polymarket US (PMUS) is a separate, CFTC-regulated U.S. exchange, distinct from global Polymarket and from Kalshi. It runs under a different regulatory framework and a different data source.

One terminology note trips up newcomers: platforms name their data tiers differently. What Kalshi calls a "series," TickerTracker treats as a market, and what Kalshi calls a "market" is an individual Outcome. We normalize all three platforms into one market → event → Outcome model so the same topic can be compared across venues. For a deeper head-to-head, see Kalshi vs. Polymarket.

What prediction markets are bad at

Taking them seriously means being candid about their limits.

Thin markets lie. A market with little volume or liquidity can show a price that reflects one or two participants rather than a crowd. The probability looks authoritative, and it isn't.
Manipulation is possible. In a low-liquidity market, a motivated actor can move the price to create a misleading headline. Depth is the defense. A heavily traded market is far harder to push.
Calibration isn't perfect. Markets show well-documented biases. The clearest is the long-shot bias, where very unlikely outcomes trade a little higher than they should. A market is a strong forecast, not an oracle.
Some questions don't fit. Markets work best on clearly defined, verifiable, time-bound questions. Vague or far-off questions with no clean resolution attract little trading and produce noisy prices.

None of this invalidates the tool. It just means a market price should be read with its volume, its open interest, and its liquidity in hand, never in isolation.

Where TickerTracker fits

TickerTracker is a volume-analytics platform for people who want to understand prediction markets rather than trade them. It unifies data across Kalshi, Polymarket, and Polymarket US into one model. That lets you see, across every venue, which questions are drawing the most money and activity, and read what that money-weighted crowd is forecasting. The lens is scale and activity. How big is this market, how busy is it, and what is it saying?

If you're new to the space, the most useful next step is simply to watch where the volume is. Browse the markets and start there.

Common questions

Is a prediction market the same as sports betting or gambling?

They overlap in mechanics but differ in purpose. A sportsbook sets odds to guarantee its own margin. A prediction market's price is set by participants and reads as a probability. For researchers and reporters, the value is in that price as an information signal, a forecast to study rather than a wager to place.

How is the probability calculated?

It isn't calculated by anyone. It emerges from trading. Because a contract pays $1 if correct, its price in cents is the implied probability. A contract at 72¢ means the market collectively prices the outcome at about 72%.

Are prediction markets accurate?

Often, especially in deep, high-volume markets, where they've matched or beaten polls on some major events. But accuracy depends on liquidity. A thin market's price can be unreliable or even manipulated, so read the price alongside the trading volume behind it.