Pick the surface that fits your runtime. All use the same Octagon API key and the same underlying research pipeline.
Latest analyzed snapshot of every active Kalshi event — model and market probabilities, edge in pp, ranked drivers, executive summary, per-outcome breakdowns.
GET /prediction-markets/events
- Build a dashboard of every event tracked by Octagon
- Surface the highest-edge events across the universe
- Pipe model probabilities into your own trading model
Read the Events API docs → Twenty typed endpoints turning the Kalshi universe into a queryable database — keyword + semantic search, thematic and behavioral clusters, pairwise correlations, Kelly sizing, basket construction, backtesting.
GET /kalshi/marketsGET /kalshi/markets/similarGET /kalshi/clusters · /behavioral-clustersPOST /kalshi/markets/correlationsPOST /kalshi/markets/edgePOST /kalshi/baskets/build · /size · /backtest · /validate · /candlesGET /kalshi/markets-with-edge
- "Find 5 uncorrelated bets on macro themes" — one HTTP call
- Backtest a basket: total return, Sharpe, max drawdown
- Rank markets by model edge, filtered by category or volume
Read the Kalshi Search & Baskets docs → Model Context Protocol server exposing Octagon's research, market data, and basket-construction tooling to MCP-capable clients (Cursor, Claude Desktop, ChatGPT).
MCP over stdio + HTTP transport
- Connect ChatGPT / Claude / Cursor directly to Octagon research
- Let agents discover and call Octagon tools without custom integration
Read the MCP Server docs → TypeScript SDK for building custom research agents on top of Octagon's data and orchestration layer.
npm install @octagon/agents-sdk
- Build a domain-specific research agent
- Compose Octagon tools with your own model + memory
Read the Agent SDK docs → Interactive web playground to try every endpoint with real data — no install required.
- Prototype API calls before writing code
Playground ↗