Prediction-market news — the crowd tells it first.
Crowdtells is a free, independent news service that treats prediction markets as a signal of what matters. It reads the money on Polymarket and Kalshi — where people bet real money on real-world outcomes — to find the questions the crowd cares about right now, then briefs each one with grounded, cross-source reporting.
Every 30 minutes we score active markets by how newsworthy they are: trading volume and 24-hour activity, resting liquidity and open interest (how much conviction is behind a position), how sharply the odds have moved, and how contested the question is. The top stories are selected with a diversity pass so one topic or one platform doesn't crowd out the rest.
Each briefing is generated by an AI model (Groq) grounded in real, recent news coverage retrieved for that market. We report what the sources agree on, where they diverge, and how outlets frame it — and a "Market lens" on how the crowd's money lines up with the press. Market figures (odds, volume, the cross-platform gap) are never written by the model; they're filled in from the live market data at render time, so the numbers in the prose always match the market and can't be fabricated. Cited sources are linked on every story.
Prediction-market odds are prices set by people betting real money, and they read as probabilities. "Donald Trump 63%" means the crowd, in aggregate, is pricing a 63% chance of that outcome — not a guarantee, and not our opinion. The number moves in real time as money comes in, which is exactly what makes a sharp swing newsworthy. When the same question trades on both Polymarket and Kalshi, we show the gap between them — a sign of how settled (or contested) the crowd's read really is. Treat it as a fast, money-backed signal to read alongside the reporting, not as financial advice.
Crowdtells is an independent project built and operated by Samuel Jo (source on GitHub). It is not affiliated with Polymarket, Kalshi, or any outlet it cites.
Questions, corrections, or press: sompiup@gmail.com.