Blog
Data essays and tutorials from the team building the statistical truth layer for prediction markets — 3.3M wallets, 84 skill metrics, every number traceable on-chain. RSS
Yes. A Polymarket backtest needs past trades, a fill model and a fair baseline. This is what a backtest must get right, what it cannot know, and how to run one from a plain-language question.
Copying a profitable Polymarket trader does not hand you their result. The price you can still get, the fills the order book allows and how the wallets were chosen decide it. Here is how to test it first.
Profit ranks bankrolls and win rate ranks price preferences. To rank Polymarket traders by skill, score calibration, edge over the price paid and risk, then check the sample size.
A wallet's profit and win rate cannot tell skill from luck. Compare how often its side won with the prices it paid, over enough resolved positions. Here is the check, step by step.
There is no good win rate on Polymarket without the price paid. A 95% win rate at 95¢ is break-even before fees; a 30% win rate at 20¢ beats the odds. Here is how to read the number.
Newsletter
The weekly Skill Report
Who moved, where the skilled money diverges from the crowd, and the calibration leader of the week. One email, no spam, unsubscribe anytime.