What is a prediction market?

Ask ten people whether a particular candidate will win an election and you get ten opinions, all of them free. Ask them to put something at stake behind that opinion and you get something different: a number that reflects what they actually believe, weighted by how strongly they believe it.

That is a prediction market. Each question resolves to a plain yes or no — the candidate wins or they do not — and participants take one side or the other. The proportion backing each side becomes a price, and that price reads directly as a probability. A question trading at 68% means the crowd collectively rates that outcome as roughly 68% likely.

Why the number tends to be good

Prediction markets have a long track record of outperforming pundits, and often polls, on questions with a clear resolution date. Three things drive that.

Skin in the game. Talking is free; committing is not. Someone who has to back an opinion thinks harder about it than someone posting a take.

Continuous updating. A poll is a snapshot from the week it was taken. A market reprices the moment news lands, which means the number in front of you already reflects this morning’s developments.

Self-correction. If a price is obviously wrong, the people who notice have an incentive to push it back. Confident wrong opinions get expensive.

None of this makes markets infallible. Thinly participated questions move on very little activity, and markets inherit the crowd’s blind spots — they were wrong about plenty of famous events, in the same way a weather forecast is sometimes wrong. Treat the number as a well-informed estimate, not an oracle.

What questions actually look like

A good question is unambiguous and has a date. “Will this policy be successful?” is not a market — nobody can say when or how it settles. “Will this bill be signed into law before 31 December?” is, because on 1 January the answer is a matter of public record.

In practice they cluster around a handful of areas: elections and politics, crypto prices, sports outcomes, economic releases, and awards and entertainment. What they share is a public, checkable answer arriving on a known date.

Multi-outcome questions

Not everything is yes-or-no. “Who wins the championship?” has a field of contenders, so it is handled as a set of linked yes-or-no questions — one per contender — whose probabilities sum to roughly 100%. Reading one is the same skill: each contender has a number, and the numbers are shares of the same whole.

Where PolyGround fits

PolyGround is a prediction market game rather than a market you put money into. You get the same question format and the same live probability swings, but you make your calls with virtual coins that have no cash value and cannot be bought, cashed out, or exchanged for anything. Being right earns XP and leaderboard position; being wrong costs coins you did not pay for.

If you want to learn to read these markets, that is a genuinely useful way to do it — the feedback is real even though the stakes are not. Next: how the mechanics work, or how to read the odds.