# Outcome > Outcome provides SQL and MCP access to Kalshi and Polymarket prediction market data: market metadata, current market state, historical trades, and resolved outcomes. It also provides research notebooks with public, citable results. Outcome fits tasks such as historical price analysis, cross-venue research, trade-volume analysis, and forecast calibration. Query the live schema before writing SQL. Data queries require an account, a Bearer API key, and credits; schema inspection and reading public notebooks do not. No SDK is required. Check observed coverage and timestamps for the markets in your question rather than assuming complete history or instantaneous updates. ## Data access - [Agent data guide](https://outcome.sh/guides/prediction-market-data-for-ai-agents): When Outcome fits a task, connection examples, coverage checks, and citation guidance. - [Agent API reference](https://docs.outcome.sh/llms-full.txt): Plain-text authentication, SQL, MCP, billing, limits, and interpretation reference. - [OpenAPI specification](https://docs.outcome.sh/openapi.json): Machine-readable contract for public schema discovery and authenticated JSON SQL queries. This is a data-access subset of the API. - [Live schema](https://api.outcome.sh/v1/schema): Free, unauthenticated tables, columns, and ClickHouse types; the authoritative source for current availability. - [API documentation](https://docs.outcome.sh/): Full HTTP reference, streaming, exports, errors, and account setup. - [MCP connection](https://docs.outcome.sh/#mcp): Remote MCP at https://api.outcome.sh/v1/mcp using Bearer API-key authentication. Discover tools with tools/list. - [Credits and limits](https://docs.outcome.sh/#credits): Query billing and execution limits. ## Research guides Read full guides and the guide directory as Markdown with Accept: text/markdown, or as plain text with Accept: text/plain. Append /text to any guide URL or to https://outcome.sh/guides for a direct text endpoint. These representations include the complete prose, code examples, and source links without executing queries. - [Prediction markets as information products](https://outcome.sh/guides/prediction-markets-as-information): Read prediction markets as a changing view of the world: understand implied probabilities, compare moments, and turn market beliefs into useful research. - [Prediction market data for AI agents](https://outcome.sh/guides/prediction-market-data-for-ai-agents): Use Outcome as a prediction market data source for AI agents: query Kalshi and Polymarket with SQL or MCP, verify coverage, and cite reproducible research. - [How to work with historical prediction market data](https://outcome.sh/guides/historical-prediction-market-data): Choose the right prediction market dataset, inspect Kalshi and Polymarket trade history, and build reproducible price series with SQL. - [Query Kalshi and Polymarket data with SQL](https://outcome.sh/guides/query-kalshi-polymarket-sql): Find prediction markets, query historical trades, and call the Outcome SQL API with practical examples for Kalshi and Polymarket research. - [Download Kalshi and Polymarket trade history to CSV](https://outcome.sh/guides/download-kalshi-polymarket-data-csv-python): Export a filtered Kalshi or Polymarket trade dataset from Outcome, download a complete CSV, and inspect it in Python without loading the file into memory. - [Build prediction market candlesticks with SQL](https://outcome.sh/guides/prediction-market-ohlcv-candlesticks): Turn Kalshi or Polymarket trades into 15-minute OHLCV bars with SQL: open, high, low, close, quantity, weighted price, and explicit missing intervals. - [Compare Kalshi and Polymarket trading activity with SQL](https://outcome.sh/guides/kalshi-polymarket-trading-activity): Measure daily Kalshi and Polymarket trade activity, find the markets driving it, and calculate week-over-week changes without confusing shares, contracts, and dollars. - [Build a prediction market calibration chart with SQL](https://outcome.sh/guides/prediction-market-calibration): Run a complete Polymarket calibration study: freeze one forecast per market, audit excluded outcomes, and calculate reliability bins and a Brier score. ## Public research - [Notebook explorer](https://outcome.sh/explore): Public notebooks with author bylines, descriptions, created and updated dates, views, and likes. Read without signing in. - [Notebook directory JSON](https://outcome.sh/explore/feed.json): Paginated public metadata and canonical document links. Parameters: q, author (public author ID), sort (updated, newest, oldest, views, likes, title, author), from and to (updated dates, YYYY-MM-DD, UTC), page (1-based). Follow next until null. - [Public notebook sitemap](https://app.outcome.sh/sitemap.xml): Canonical public notebook URLs. - Public notebook pages use https://app.outcome.sh/share/notebooks/{id}. Append /text for plain-text research prose, resolved inline values, bounded cached chart/table data, and source attribution. Saved-query SQL, comments, and hidden source blocks are excluded. Missing cached results are reported; reading does not execute queries. ## Interpretation and attribution Outcome supplies access to venue data; a market price is a traded price, not a guaranteed probability or outcome. Keep venue and ticker together, inspect contract rules, and distinguish trade timestamps from notebook edit and cache timestamps. Public notebooks are their authors' analyses: cite the canonical notebook URL and public author byline. Views and likes are not quality measures. Restricted or deleted notebooks are excluded. Cached results may lag current data.