Why prediction markets and polls disagree
A poll asks how people would vote today. A market prices who wins on the day.
Polls and markets answer different questions, so they often disagree without either being broken.
What a poll measures
A poll asks a sample of people how they'd vote if the election were held now. It comes with a margin of error, it's a snapshot of one moment, and it doesn't tell you who wins. A candidate up 3 points in a poll with a 3-point margin of error could easily be behind.
What a market measures
A market prices the final outcome. Traders fold in the polls, but also turnout, the time left, how the rules count votes, and anything else they think matters. That's why a candidate up 3 in the polls can be priced at 75% to win: winning by 1 still counts as winning.
Why they can split
Different questions. "Who is ahead" and "who wins" aren't the same, especially in systems like the US Senate, where control depends on a handful of close states. See every race on the midterms page.
Timing. Markets move in minutes. Polls take days to field and publish.
Money. A big trader can push a thin market away from where the evidence points, at least for a while. Polls have their own biases, like who answers the phone.
Read them together
When polls and markets agree, that's a stronger signal than either alone. When they split, the gap is the story: ask what the market might know, or might be getting wrong. Next: how markets settle.