About Market Guy

What this is

Market Guy is a research workbench for prediction markets. It pulls live contracts from Polymarket and Kalshi, adds the context a price alone does not carry — the underlying asset, comparable historical base rates, how the wallets with a demonstrated edge are positioned — and produces cited research reports you can check line by line.

The product exists because a prediction market price is a number without a story. It tells you what the crowd currently charges for an outcome and nothing about why, what would change it, or how often something like it has happened before. Filling that gap is the whole job.

Who it is for

People who want to understand a market before they take a position in it, and people who follow prediction markets as a signal about the world without trading at all. Journalists, analysts and the merely curious get the same research surface as anyone else.

It is explicitly not a tipster service. There is no signal feed, no alerts, no picks. If you are looking for someone to tell you what to buy, this is the wrong product and we would rather say so here than after you have paid.

Where the numbers come from

Market prices, order books and trade history come from the venues themselves. Underlying asset data — spot prices, three years of daily closes, perpetual funding and open interest — comes from public exchange endpoints. Crypto market capitalisation and supply come from CoinMarketCap. The research reports are generated with Perplexity's Sonar models against live web sources, and every factual claim carries a citation you can open.

The full list, including how often each source refreshes and what happens when one fails, is on the data sources page. The research method and its limits are on the methodology page, and the model card names six failure modes rather than a capability list.

What we do not do

We do not give financial advice, and nothing on this site is a recommendation to buy, sell or hold anything. We do not operate a market, hold customer funds, or execute trades. We are not affiliated with Polymarket or Kalshi.

We also do not publish a number we cannot support. Where the data is too thin for a claim, the product says so instead of estimating — that rule shows up as blank fields, missing base rates and scans that come back empty, and all three are deliberate.

How we measure ourselves

Every AI analysis is logged with the probability it stated and the market price at the time, and every smart-money verdict with the price at that moment, so both can be scored against outcomes later rather than remembered selectively. Those scores are used internally to improve the method; the rules they follow are on the signal scoring page. What we publish is how accurate the markets themselves have been, in the market accuracy study.