Prediction markets are supposed to give you probabilities, not outcomes. The real world is uncertain, and @Polymarket quantifies that uncertainty with probabilities. So if it's not "Polymarket predicted X and X did/didn't happened"—if it just outputs probabilities—how do you measure how predictive Polymarket actually is? Statisticians already have a tool for this, called the Brier Score. Brier Scores basically tell you: "when you say an outcome is ~60% likely to happen, do those things happen 60% of the time?" Same for 10%, same for 90%, and so on. Here's what that looks like for Polymarket, @Dune dashboard by @alexmccullaaa 👇 For Brier Scores, smaller is better. Anything below 0.1 is generally fantastic. (Sports betting lines tend to be around 0.2). Polymarket's Brier Score is 0.056, which means it is extremely accurate at surfacing correct probabilities. Takeaway: Polymarket is very accurate, but it's subtle what that means and how it's measured.