01.Informational edge
You know something the market has not priced — a domain you work in, a data source you read that others do not, or simply having done the reading. This is the most durable kind, and also the rarest, because on well-traded markets a lot of people have done the reading too. It decays as soon as the information becomes general.
- Strongest in specialist domains where few traders have background.
- Decays the instant the information becomes widely known.
- Ask honestly what you know that a motivated generalist could not find in an hour.
02.Attention edge
The market is not wrong because information is hidden but because nobody has bothered to look. Long-tail markets, obscure outcomes in a multi-outcome board, and questions that were listed and then forgotten all drift from fair value simply through neglect. This is the most accessible edge for an individual — and it comes packaged with the thin liquidity that made it possible.
- Long-tail outcomes on multi-outcome boards are the classic case.
- The same neglect that creates the mispricing makes it hard to size into.
- Sizing, not identification, is the binding constraint here.
03.Structural edge
The prices themselves are internally inconsistent: a binary's two sides summing away from a dollar, related markets implying contradictory probabilities, or the same event priced differently on two venues. This kind does not require you to know anything about the world, only to notice arithmetic. It is also the most competed, because bots find it faster than people do.
- Cross-outcome sums that do not add to a dollar.
- Related markets implying probabilities that cannot both be true.
- Cross-venue gaps on equivalent events — mind the differing resolution criteria.
04.Testing whether you actually have one
Before sizing, write down your probability and why. If your number is close to the market's, you have no edge regardless of how confident you feel. If it is far away, ask what the people on the other side know that you do not — because someone is taking that trade, and the burden is on you to explain why they are wrong.
- State your probability numerically before looking at the market price.
- A small gap is not an edge; it is noise plus wishful thinking.
- If you cannot say why the other side is wrong, you may be the other side.
05.What is not edge
Strong feelings about an outcome, agreeing with a confident poster, or wanting a particular result. Each of those produces conviction without producing information, and prediction markets are unusually good at extracting money from that combination — because the person on the other side is often being paid precisely to take it.
- Conviction is not information.
- Wanting an outcome actively degrades your estimate of it.
- Following a confident stranger is copying their bias, not their edge.