Octagon Review

Octagon is an AI-powered market intelligence platform that sits on top of prediction market exchanges like Kalshi. Rather than just showing you the current price, it generates independent probability models for market events in real time, then highlights the gap—called edge—between what its AI thinks should happen and what traders are actually pricing in. Every claim in the research is cited to source, and models update around the clock as new information surfaces.

The features that matter

• Continuously updated research reports: One report per active market, with model probability, current market price, and edge calculation. All refreshed as news breaks and new data arrives.
• Real-time edge detection: The platform shows you exactly where the model & market disagree—flagged in reports, a live news feed, and via API endpoints ranked by expected return.
• Cited, auditable analysis: Every prediction is stamped with sources. You can verify the reasoning behind each model probability rather than accept it as a black box.
• Four surfaces, one engine: Access the same underlying intelligence through research reports, a signals news feed, a terminal-based trade CLI, or a REST API with semantic search and thematic clustering.
• Coverage at scale: Currently tracking 1,000+ Kalshi markets and 54,000+ contracts with plans to expand to Polymarket.

How it stands out

The critical distinction here is independence. Kalshi and Polymarket are exchanges—they show you order books and trade volume. Octagon is the research layer on top. Its multi-agent AI system reads news sources, financial data, and market signals to build its own probability estimate for each event, then compares that to live prices. If Octagon's model says an event has a 46% chance but traders are pricing it at 58%, you see that 12-point gap flagged immediately. Whether that's edge you can exploit depends on your conviction in the model, but the platform makes the disagreement explicit and traceable.

Honest trade-offs

The accuracy of Octagon's edge depends entirely on the quality of its probability models. A cited source is transparent, but transparency doesn't guarantee correctness—markets sometimes price things accurately that models miss. You're also limited by coverage: right now it's Kalshi-focused with Polymarket coming later. And if you want to actually trade on the edge you find, you'll need to move to the CLI or API; browsing reports alone won't execute positions. The platform works best for traders who want research depth alongside execution, not for passive observers.