Election Prediction Markets: How They Work And What Their Prices Mean
Election prediction markets are the reason most people have heard of prediction markets at all. Every campaign season, a set of numbers starts moving on exchange screens, gets quoted on television next to the polls, and gets argued about as if the numbers themselves had picked a side. They have not. An election market price is an instrument reading, a live probability estimate produced by people risking real money, and the most famous nights in the category's history are the nights that reading was misread. By the end of this piece you will be able to read a 30¢ contract the way the market means it, and you will understand why the elections where the trailing side won were not proof the instrument was broken. One thing before we start: this article is about the instrument. It takes no side, names no candidate, and works the same no matter how you vote.
The Quick Answer
An election prediction market is an exchange where traders buy and sell contracts that pay $1 if a specified election outcome happens and $0 if it does not. The price, quoted in cents, is the market's live estimate of the outcome's probability: a 30¢ contract means the crowd, with money at stake, puts the chance near 30%, which also means it expects that outcome to lose more often than not. How the contracts are built, what a price can and cannot tell you, how these markets stacked up against polling, and why a 30% outcome actually happening does not make the market wrong — all of it is below.
The Contract: One Dollar, Split By An Outcome
Every election market is built from the same part: a binary contract on a named, verifiable outcome, such as which party wins a chamber of Congress or which candidate wins a race. The contract trades between $0.01 and $0.99 while the question is open, then settles at exactly $1 for the side that happened and $0 for the side that did not, against an official certified result named in the market rules. You can take either side. If you think YES at 30¢ is too cheap, you buy it; if you think it is too expensive, you can sell it instead, collecting the 30¢ now and owing the full dollar only if the outcome happens, which is the same position as buying NO at 70¢. The mechanics are the same as any other event contract, and our guide to how prediction markets work walks through them from zero.
Two structural details do most of the work:
- There Is No House. An exchange matches buyers with sellers and charges a fee; it does not set the number or take a position against you, which is the core reason an exchange is not a sportsbook. The price is simply the last level where a buyer and a seller agreed, so it moves the moment the crowd's estimate moves.
- The Board Is A Distribution. A race with several possible winners lists one contract per outcome, and those prices have to sum to roughly a dollar, because exactly one of them will be worth a dollar when it ends. That constraint turns the board from a menu of numbers into a probability distribution you can read at a glance.
In the United States, contracts like these trade on regulated venues, including Kalshi, which operates as a CFTC-regulated exchange.
The structure, in other words, is simple. The number it produces is where the misreadings happen, so that number deserves its own section.
What A Price Means As A Probability
A price in an election market is a probability wearing a dollar sign, and the cleanest way to see it is to run one trade end to end.
A Worked Example: $30 On A 30¢ Contract
| The Trade | The numbers |
|---|---|
| Buy 100 YES Contracts At 30¢ | Cost: $30 |
| The Outcome Happens | Contracts settle at $1 each: collect $100, a $70 profit |
| The Outcome Does Not Happen | Contracts settle at $0: the $30 is gone |
| Break-Even Rate | 30%, which is why the price reads as the probability |
For that trade to break even over time, the outcome needs to happen at least 30% of the time, so a trader paying 30¢ is asserting the true chance is above 30%, and the seller taking the other side is asserting it is below. The market price is where those two assertions currently balance. If you already think in American odds, the translation is mechanical, and Kalshi odds explained covers it in detail:
| Price | Implied probability | Sportsbook-odds equivalent |
|---|---|---|
| 10¢ | 10% | +900 |
| 30¢ | 30% | +233 |
| 50¢ | 50% | +100 |
| 70¢ | 70% | -233 |
| 90¢ | 90% | -900 |
The row worth staring at is 70¢. A 70¢ contract describes an outcome that fails three times in every ten, which is roughly the rate at which an NFL favorite of about six points, priced near -240 on the moneyline, loses. Nobody who watches football calls a six-point favorite a finished game, yet a 70-30 election market gets talked about as if it were already decided. The price never said that.
There is one wrinkle the table hides. A sportsbook price carries the book's vig: quote -110 on both sides of a game at DraftKings or FanDuel and the implied probabilities sum to about 104.8%, which works out to roughly a 4.5% hold for the book, so the honest comparison to an exchange requires a de-vig to the no-vig fair price first. An election ladder that sums to roughly a dollar is already close to a no-vig book, which is one reason researchers like quoting exchange prices as probabilities in the first place.
It is just as important to hear what the price is not saying. A 70¢ contract is not a projection that anyone wins 70% of the vote, and it says nothing about the margin. It is not a poll, either: a poll asks a sample of people what they intend to do, while a market price is a claim about the final certified outcome, filtered through everyone willing to back their read with money. And it is not a certainty about anything. It is a probability, which means it is a statement about frequencies, and that is exactly the property the famous nights got wrong.
Why A 30% Outcome Happening Does Not Make The Market Wrong
A 30% estimate is a claim that, run ten times, the outcome shows up around three of them. An election is not run ten times. It resolves once, and one resolution cannot grade a probability. When a contract priced at 30¢ settles at $1, there are two live explanations: the market underrated the outcome, or a three-in-ten shot came in, the way three-in-ten shots do, three times out of ten. A single night cannot tell you which one you watched.
This is not a technicality; it is the whole discipline of reading these markets. More than one high-profile election has ended with the side priced in the 20s and 30s winning, and those nights are still cited as proof that election prediction markets failed. Flip the logic around and the accusation collapses: if outcomes priced at 30¢ never happened, the markets really would be broken, because those contracts would be systematically overpriced and sellers would eventually trade that mispricing away. Long shots that sometimes land are what correctly priced long shots look like.
A probability is graded on a sample, never on a single night. A 30¢ contract that settles at $1 is either an underpriced outcome or a three-in-ten shot doing what three-in-ten shots do. One election cannot tell you which, and anyone declaring the market broken from one result is reading the instrument wrong.
The honest test of a probability is calibration over a sample: collect every contract that traded near 30¢, let them all settle, and check whether roughly 30% resolved YES. That takes a lot of markets and a lot of patience. We hold our own trading to exactly this test in a much smaller arena, temperature markets, where we publish every settled position on our Kalshi weather markets hub, and the honest reading there is that it takes on the order of a thousand settled contracts before a small gap between implied and realized probability separates from noise. If we do not let ourselves judge our own probabilities on a handful of nights, one election night cannot judge a market.
How Election Markets Compared With Polling
The oldest evidence here predates every modern platform. The best-known research project is the Iowa Electronic Markets, a small real-money election market the University of Iowa launched in 1988, and the broad finding from the academic work built on it and projects like it is that market prices held up well next to polling averages as election forecasts, with one structural advantage: they adjust to new information continuously instead of waiting for a survey's fieldwork window to close. That result is less mysterious than it sounds. A market is not an alternative to polling; it is a machine for aggregating polls along with everything else, including turnout models, early-vote data, historical base rates, and the occasional trader who simply knows a state better than the national conversation does. It is the same aggregation logic we watch settle against a thermometer every day in our temperature markets, just pointed at a slower question.
The same relationship explains the failures. Traders read polls, so when polling misses in a correlated way across states, markets tend to miss with it, and the famous upset nights embarrassed both instruments at once. A market is a poll of everyone trading on the polls, sharpened by money and updated continuously, not an oracle with independent access to the truth. We looked at the broader evidence, including where the crowd earns its reputation and where it does not, in are prediction markets actually accurate.
What The Price Cannot Tell You
Reading the number as a probability is the main skill. The second skill is knowing when the number is speaking softly. Election markets attract heavy attention on the headline race and far less everywhere else, and a lightly traded contract has a thin order book, where one modest order can move the printed price several cents without any information arriving. Watching where money is actually resting is a more honest read than watching the headline print, and that order-flow view of exchange and prediction-market money is what the OddsShopper Liquidity Tool tracks. The same contract can also trade at different prices on different venues at the same moment, which is less an opportunity than a warning about how soft a single print can be; the price gaps across exchanges are a lesson in their own right. On top of that, exchange fees shave a little from every trade, and how Kalshi's fees work shows the fee is largest right in the middle of the range, where election contracts spend their most contested months.
The sharpest warning sits at the ends of the board. Late in a cycle, one side of a market drifts into the 90s, and the other side sells for a few cents, and both start to look like the safest trades on the board. They are not. Selling an unlikely outcome collects a small premium and risks most of a dollar, so a single loss erases the premiums from something like thirty or forty winning trades, and that arithmetic, not any hit rate, is what makes position sizing the entire game. We wrote up the math in when you sell a long shot, one loss costs many wins, and election markets are where the temptation runs strongest, because the crowd's certainty peaks exactly when the remaining tail is priced smallest.
Where These Markets Run
Availability is a moving map, and election contracts have a particular history on it: they reached regulated US exchanges only after Kalshi beat the CFTC in federal court over whether its election contracts could list at all, and the rules around who may trade them can differ by venue and by product. In the United States, election contracts trade on CFTC-regulated exchanges such as Kalshi, while other venues operate under different structures and different rules; are prediction markets legal covers the landscape, and our looks at whether Kalshi is legit and Kalshi vs Polymarket cover the two platforms readers ask about most, with a fuller rundown in our prediction market platforms guide. One disclosure belongs right here rather than in a footer: we have no affiliate or commercial relationship with Kalshi, and we do carry sign-up offers for some other prediction-market and betting platforms, so read any platform comparison we publish with that incentive in mind.
Election Prediction Markets FAQ
Is an election market price the same thing as a poll number? No. A poll estimates current voter intention from a sample; a market price estimates the probability of the final certified outcome, using polls as one input among many. A candidate can lead a poll 52-48 while trading at 80¢, and both numbers can be right, because they measure different things.
What does it mean when a contract trades at 95¢ before the result is official? The market considers the outcome nearly settled but not settled. The last few cents exist because rare things happen, and the trader selling that residual doubt is risking most of a dollar to collect pennies. A 95¢ price the night of a count is a strong read, not a certified result.
Can a big trader push an election market price around? They can move the printed price, especially in a thin book. Holding it somewhere the evidence does not support is much harder, because a distorted price pays everyone else to trade it back, and settlement against the certified result has the final vote.
What happens if a race is contested or takes days to call? The contract stays open until the outcome named in the market rules is official. A recount, a certification fight, or a race that sits uncalled delays settlement rather than voiding it, so you can be left holding an unsettled position well past the night itself. Reading a market's resolution rules before you trade it is the cheapest protection these markets offer.
Are election prediction markets legal in the US? On regulated exchanges, election event contracts are available to eligible US users, with rules that differ by venue and product. The details shift enough that they get their own page: are prediction markets legal.
An Instrument, Not An Oracle
Read back the thread and the famous nights look different. A price is a probability: 30¢ claims three-in-ten, and three-in-ten things happen. A market is an aggregation machine that digests polls rather than competing with them, so it inherits polling's correlated misses along with its information. And a probability can only ever be graded on a sample, never on a single evening, which is the same discipline we apply to our own public log of temperature-market trades, where the losses are published next to the wins precisely because a handful of results proves nothing. That is the whole instrument. It will not tell you who should win, and it was never wrong just because an underdog got there. If you want to see probability-first thinking applied with real stakes in a different arena, our free expert picks are open to read, no account needed.
Disclosure and fine print. Stokastic trades Kalshi weather markets and holds positions in them; where a settled position is described in this series, we were the seller. We have no affiliate or commercial relationship with Kalshi; we do carry sign-up offers for some other prediction-market and betting platforms. Kalshi contracts are CFTC-regulated event derivatives traded on a designated contract market, not sportsbook wagers, and a position can lose its full value. Selling an unlikely outcome collects a small premium and risks most of a dollar, so one loss can erase the premiums from thirty or forty wins. 18+, available where Kalshi operates; the risk of loss is real and, on the side we trade, individually large. This series is an open research log of a strategy we have not proven, and its sample is still far too small to judge. Nothing here is trading advice, and nothing on this page is a pick, a play, or a recommendation about any election outcome.
