A poll tells you how much support a candidate has today. A prediction market tells you how likely traders think that candidate is to win. They are different measurements, and reading one as if it were the other is the most common mistake people make during election season.
What a poll measures
A poll interviews a sample and projects it to the electorate. That covers horse-race numbers such as Iowa Senate polls or Michigan Senate race polls, head-to-heads in a primary, and approval ratings for a sitting official.
- Margin of error applies to each candidate, so a 2-point gap is often inside the noise.
- House effects mean pollsters differ from each other in consistent ways.
- Approval rating is not vote share. An approval rating says how people judge the job being done, not how they would vote against a named opponent.
- Averages beat single polls. One outlier moves headlines more than it should.
What a prediction market measures
On a prediction market, a YES share pays $1.00 if the event happens and $0 if it doesn't. A price of 68¢ reads as an implied probability of about 68%. The price updates with every trade, so it reacts to a debate, an indictment or a new poll within minutes.
| Poll | Prediction market | |
|---|---|---|
| Reports | Vote share or approval | Probability of an outcome |
| Updates | When a new poll is released | Continuously |
| Who speaks | A sample of voters | People risking money |
| Typical weakness | Sampling, non-response, house effects | Thin liquidity, large traders |
Why the numbers don't match
A candidate ahead 52-48 in the polling average is not "52% likely to win". If that lead is stable and polling error is usually around three or four points, the lead holds up most of the time, and a market might price the race near 70¢. The poll gives the margin; the market turns the margin and its uncertainty into a probability.
It works the other way too. A candidate at 40% in a three-way primary can be a clear favourite, while one at 45% in a two-way general election can be a clear underdog.
How to use both
- Start with the polling average to see where the race stands.
- Read the market's exact question. "Wins the seat" and "party controls the Senate" are separate markets.
- Check liquidity. Look at volume and the spread before trusting a price.
- Treat disagreement as a question. If polls and price diverge, ask what one side knows that the other doesn't.
For the mechanics of election contracts, see Polymarket political markets and researching 2028 presidential election predictions.
A note on risk and access
A probability is not a promise: 80¢ loses one time in five, and the buyer loses the full price paid. In the United States the international Polymarket site is close-only; Americans use the CFTC-regulated Polymarket US app, where it is available in their state.
Official source: Polymarket docs: Markets and events.