What Prediction Markets Price That Polls Cannot

A poll asks people what they believe. A prediction market asks people what they will bet. Those are not the same question, and the gap between them is the entire edge.

Treat a market price as a probability and a poll as a guess. One has skin in it. The other does not.

Stakes change the answer

When an answer costs nothing, people give the answer they want to be true. They signal. They vote for their team. They round to the story they already hold.

Put a price on the answer and the behavior changes. A trader who is wrong loses capital. That single fact strips out most of the noise that pollutes survey data. What is left is closer to a real expectation, weighted by how much conviction each participant is willing to fund.

This is why a thin market can still beat a large poll. A poll of 2,000 people with no stake is 2,000 opinions. A market with 200 funded participants is 200 forecasts that each cost something to be wrong about.

The number is a probability, not a headline

A contract trading at 0.62 is not a prediction that an event will happen. It is the market clearing price for a claim that pays 1 if it happens and 0 if it does not. Read it as a probability and the whole instrument gets useful.

That reframing does real work:

• A move from 0.40 to 0.55 is not “the market flipped.” It is a 15 point repricing of probability, and the interesting question is what information moved it.
• A price stuck at 0.50 in a liquid market is not indecision. It is a genuine coin flip, and most narratives insisting otherwise are selling something.
• A price that gaps and holds is the market telling you new information arrived and stuck.

Translation: stop reading the level and start reading the change. The edge is in the move and the reason behind it, the same way it is when you read past a revenue print to the bookings underneath it.

Where the edge actually is

The edge is not in being smarter than the market on the headline question. By the time a contract is liquid, the obvious view is priced. The edge is in three quieter places.

1. Mispriced tails. Markets are bad at very low and very high probabilities. A 0.03 contract that should be 0.08 is a better trade than a 0.50 that should be 0.52, because the payout structure rewards the correction far more.
2. Slow information. Some markets reprice news in minutes. Others take a day because the participants are not watching. The lag is the trade.
3. Structural sellers. When one side of a market is dominated by people hedging rather than forecasting, the price drifts away from fair value and stays there. Find the hedger and you have found the edge.

None of that requires a better opinion. It requires reading the market as a system of incentives and finding the seams where the incentives do not point at accuracy.

A market is not wise because the crowd is wise. It is wise because being wrong is expensive. Remove the cost and you remove the wisdom.

How I use them

Prediction markets are an input, not a strategy. They sit alongside the rest of the work.

The most useful thing they give me is a calibration check. When my read on an outcome diverges sharply from a liquid market, one of two things is true. Either I am early and the market has not caught up, or I am wrong and the market is telling me so. Both are worth knowing before sizing anything.

The second use is faster than any news feed. A market that gaps before the story breaks is the story breaking. Watching the right contracts is watching informed money move in real time.

I do not bet outcomes I have no edge on. Most contracts are efficient. The discipline is the same as it is in equities. No trade without a thesis. No thesis without a reason the price is wrong. Most days, the answer is that it is not wrong, and the right move is to do nothing.

The manifold view

Any single contract is noise. A close election, a rate decision, a binary that resolves and disappears. Up close it is just gambling with extra steps.

Pull back and the picture is different. Across hundreds of resolved markets, the prices were calibrated and the polls were not. That is the structure. Stakes produce better information than surveys, consistently, because the cost of being wrong does the filtering that good intentions cannot.

Flat in any one contract. Curved across the sample. Read the change, not the level.

Still reading the structure under the noise.

Neuralmanifold · +35.0% YTD · 16 positions · 10.3% cash