Guide 6 · Markets and polls
Prediction markets vs polls: which forecasts better, and when
A poll measures what a sample of people say; a market prices what traders think will happen. Markets update instantly and reward being right, while polls are transparent about method and sample. Neither is reliably better: markets can be thin or skewed by money, and polls can miss who actually turns up.
They answer different questions
A poll asks, “Who will you vote for?” A market asks, “Who will win?” The second question lets a trader combine polls, turnout models, news and gut feel into one number. That is an advantage when information is scattered, and a weakness when traders mostly copy the same headlines.
Side by side
| Aspect | Poll | Prediction market |
|---|---|---|
| Incentive | No stake in being right | Money on being right |
| Who takes part | A designed sample | Whoever chooses to trade |
| Updates | Days or weeks apart | Continuously |
| Uncertainty | Stated as a margin of error | Implied by the price |
| Main weakness | Sampling and turnout errors | Thin trading, skew from large traders |
When markets tend to do well
- When many informed people trade and the market is deep.
- When news breaks between polls, because a price moves in minutes.
- When the question is about an outcome, not an opinion: a rate decision, a launch date, a match.
When polls tend to do well
- When the market is thin or dominated by a few large traders.
- When you need to know why, not just what: polls report who thinks what and for which reason.
- When traders share the same blind spot, such as a population that rarely trades.
How to judge either one: calibration
One missed call proves little. Collect many forecasts and check calibration: of all the events given about a 70% chance, did about 70% happen? A forecaster, a poll aggregator or a market that is well calibrated deserves trust on the next question; one that is badly calibrated doesn’t, however confident it sounds.
Use both
Read a poll for who and why, a market for how likely right now. When they disagree, ask which one has the thinner evidence behind it.