Prediction Markets

Prediction markets let participants trade contracts tied to future events, using prices to combine information into implied probabilities. Research studies how well these markets aggregate private information and how humans, language models, and multi-agent systems can forecast outcomes, resolve contracts, and turn beliefs into trades. Common approaches include probability scoring with Brier scores, ensembles and wagering mechanisms, and trading tests on Kalshi or Polymarket that account for timing, fees, and order-book execution. A central tension is that accurate forecasts do not always produce profitable trades, while AI agents have yet to consistently outperform market-price baselines under realistic evaluation.