Reading the numbers
Prediction market odds explained
The percentage is the easy part. Knowing what it does and does not tell you is where people go wrong.
A prediction market quotes a single number between 0 and 100. That number is a probability: 68% means the crowd rates the outcome as happening roughly 68 times out of a hundred. Simple to state, easy to misread.
Converting to other formats
If you are used to odds expressed another way, the translation is mechanical. A market at 25% is the same as 3-to-1 against, or decimal 4.0. At 80% it is 1-to-4 on, decimal 1.25. The general rule: decimal odds are 100 divided by the percentage.
One genuine difference is worth knowing. Bookmaker odds include a built-in margin, so the implied probabilities across all outcomes add up to more than 100%. Prediction market prices sum to roughly 100% instead, which makes them a cleaner read on what people actually believe.
The mistake almost everyone makes
A market at 30% that comes true is not a market that was wrong. Nor is one at 85% that fails. Probabilities are only judged across many predictions: if everything you called at 85% happened, your 85% was too low. A forecaster who is never surprised is miscalibrated.
The movement matters more than the level
A single number is a snapshot. The interesting information is usually in how it got there.
A question that has sat at 45% for a month is genuinely uncertain — everyone has looked at it and nobody can separate the outcomes. A question that jumped from 45% to 80% overnight is a different animal: something specific happened, and the move is the market telling you it mattered. A slow drift over weeks usually means accumulating evidence rather than a single event.
This is why every question on this site shows its history, not just its current price. The shape of the line is often the more useful half of the information.
Odds versus polls
The two answer different questions. A poll asks what people currently intend. A market asks what people expect to happen — which already folds in turnout, momentum, the remaining time, and the polls themselves.
So when a market and a poll disagree, they are not necessarily in conflict. A candidate can lead a poll while the market rates them below even, because the market is pricing in everything else it knows. Where markets have a clearer advantage is speed: they reprice within minutes of news, while polls take days to field.
Three things to check before trusting a number
How much activity is behind it. A quiet question moves on very little, so its precision is illusory.
How far away resolution is. Distant questions are systematically less sharp than ones settling next week.
What exactly it resolves on. A surprising price is very often a correctly priced question that asks something narrower than you assumed.
Getting a feel for it
Calibration is a skill and it responds to practice — but practising with real money is an expensive way to discover you are overconfident. PolyGround gives you the same questions and the same live odds with virtual coins that have no cash value, so you can find out how well you actually read these numbers before anything is at stake.
Browse live questions by topic, or start with the basics.
Try it yourself — free
PolyGround is a play-money prediction game. Real-world questions, live odds, virtual coins. No deposits, no cash-out, no crypto.
Get it free on Google Play