Live
AI forecasts, scored in public
Ask any yes-or-no question about the future. Foresea runs real-time searches to gather evidence, weighs facts via an LLM council, and calculates calibrated probabilities compared with live prediction markets.
Live contracts matched with Polymarket Kalshi
Key Intelligence Features
We run real-time evidence agents to challenge crowd consensus and identify mispriced prediction market contracts.
Unlike standard static LLMs, Foresea runs automated research agents that scrape live news databases (via news APIs and search engines) to construct an up-to-the-minute evidence brief for every forecast question.
Foresea maps its probabilities directly to order-book pricing on Polymarket and Kalshi. It flags significant contract discrepancies, helping you spot arbitrage, hedges, and mispriced markets instantly.
Transparency & Trust
Prediction markets are difficult to beat because they represent the financial stakes of thousands of active traders. To establish trust, every Foresea forecast is logged in public and scored against final real-world outcomes.
We evaluate the forecast calibration: if we group all predictions where we estimated a 75% probability, about 75% of those events should occur. We track these results transparently in our public database and track record sheets.
* Calibration values computed from actual model outcomes in the resolved track record (`track_record_live.json`). Lower Brier score indicates better calibrated predictions.