Prediction Markets
Polymarket and Kalshi prediction markets unified on Dune: per-fill trades, hourly OHLCV, resolution outcomes, all queryable in DuneSQL today.
Tables in this collection
Frequently asked questions
What is a prediction market?
A prediction market is a contract that resolves to $1 if an event occurs and $0 if it doesn't. The price at any moment is the implied probability of that outcome, set by participants trading with real capital. Dune surfaces these prices as a continuous hourly series alongside per-fill trades and resolved outcomes.
What's the difference between the Polymarket, Kalshi and Hyperliquid data?
Polymarket is an onchain order book on Polygon: every trade is a blockchain transaction and every fill carries maker and taker wallet addresses. Coverage starts in November 2022. Kalshi is a CFTC-regulated exchange with traditional accounts; trades come from Kalshi's API, so there are no wallet addresses, but each trade carries its fee, derived from Kalshi's published fee schedule. Coverage starts in June 2021. Hyperliquid HIP-4 outcome markets trade onchain on Hyperliquid, covered since May 2026. Kalshi and Polymarket share the unified prediction_markets tables; HIP-4 has its own tables.
How fresh is the data?
Tables refresh daily. Candlesticks are hourly bars, shipped with those daily updates.
Does Dune cover Polymarket V2?
Yes. Polymarket moved its exchange to V2 on April 28, 2026, and Dune covers V2 from launch. Pre- and post-migration data live in the same tables with consistent column names and market identifiers, so queries spanning the migration need no changes.
Does Dune cover Polymarket US?
Not yet. Polymarket US is on the roadmap. The current dataset covers Polymarket Global (Polygon, since November 2022), Kalshi (since June 2021) and Hyperliquid HIP-4 (since May 2026). If Polymarket US coverage is a priority for your team, flag it when booking a demo.
Can I backtest signals against historical data?
Yes. Every resolved market carries its final outcome, so the full history is usable for signal validation without additional cleaning. Kalshi coverage starts in June 2021 (40+ Federal Reserve meetings and 60+ inflation prints, each with an hourly probability path from open to settlement) and Polymarket in November 2022, all queryable from the same SQL surface.
How does Dune handle API or schema changes at the venues?
Dune absorbs upstream changes so queries built against the unified schema don't break when a venue's API shifts. Polymarket's V2 cutover on April 28, 2026 is one example: column names and identifiers stayed stable across it. The same applies to Kalshi API updates.
How do I get access?
Four ways: query in Dune's SQL editor, fetch through the Dune API, sync to your own warehouse (Snowflake, Databricks, BigQuery, S3) with Datashare, or connect AI agents through MCP. Free: five tables with no account required, three Polymarket tables (trades, market details, hourly prices) and two Kalshi aggregate reports. Enterprise adds Kalshi per-fill data, hourly open interest, Polymarket wallet positions and combo markets, Hyperliquid HIP-4, the unified cross-venue schema and ticker mapping. Book a demo to scope the right setup.
Does Kalshi have an API?
Yes. Kalshi's REST API serves public market data without an account; trading needs an API key from a Kalshi account. For research and backtesting, Dune gives you Kalshi trades, hourly prices, market details and open interest back to June 2021 in SQL, with no ingestion, pagination or rate limits to manage.
What are the Kalshi API rate limits?
Kalshi's Basic tier allows 20 read requests per second. The trades endpoint is paged, so rebuilding full history means paging call by call within that budget. On Dune, the same history is one table.
Is the Kalshi API free?
Market data endpoints are public, and the Basic rate-limit tier comes with a Kalshi account; Advanced is by request, and higher tiers are earned through trading volume. The real cost for research is engineering time: ingestion, deduplication, normalization and storage. On Dune, two Kalshi aggregate tables are free, and per-fill Kalshi data is part of Enterprise.
Does Polymarket have an API?
Yes, three: the Gamma API for market metadata, the CLOB API for order books, prices and trading, and the Data API for trades and positions. Dune combines Polymarket's API data with trades and positions decoded directly from Polygon, so every trade, position and market since November 2022 is in one SQL schema, without stitching the three together.
Do I need a Polymarket API key?
Only for private requests such as trading. Polymarket API credentials are created with a signature from your wallet and authenticate private CLOB requests; public market data needs no credentials. To analyze Polymarket markets, prices and positions on Dune, you need no Polymarket key at all.
What are the Polymarket API rate limits in 2026?
Limits are per IP over 10-second windows: 4,000 requests on the Gamma API, up to 1,000 on the Data API (trades and positions are tighter, about 200 and 150) and 9,000 on the CLOB API. Requests over the limit are throttled. Dune removes the limits from the research side: query the full history in SQL or sync it to your warehouse.
Is there one API for Kalshi and Polymarket data?
Dune covers both, plus Hyperliquid HIP-4. Kalshi and Polymarket share one unified schema, available in Dune's SQL editor, through the Dune API, via MCP for AI agents, and by Datashare into Snowflake, BigQuery, Databricks or S3. Political, macro, sports and company markets all sit in the same tables.
How do hedge funds use prediction market data as an investment signal?
Prediction markets price the probability of events that move assets: Fed decisions, earnings, tariffs, supply shocks. Funds use the hourly probability series as an event-driven signal, backtest it against resolved outcomes, and compare it with references such as CME FedWatch.
How do I map prediction markets to stock tickers?
Dune's ticker mapping matches every Kalshi and Polymarket market that prices a public company to its ticker, and labels it by event type: earnings, operating numbers, deals, legal and regulatory, listings, leadership. Join it to your holdings to see which markets matter to your portfolio.
Where can I get historical prediction market data?
Dune has Kalshi history since June 2021, Polymarket since November 2022 and Hyperliquid HIP-4 since May 2026, with every resolved outcome included. Query it in SQL, fetch it through the API, or sync it to your own warehouse.
How can a macro desk use prediction market data?
Prediction markets price the events that move rates and FX directly: Fed decisions, CPI and jobs prints, the Fed chair nomination, elections, tariffs and government shutdowns. The Fed chair nomination market alone traded over $200 million across candidates. Macro desks get the full probability distribution for each event, not just a consensus. Kalshi and Polymarket priced the April FOMC "no change" about 20 cents above CME FedWatch for six weeks before converging. Kalshi history starts in June 2021, covering 40+ Fed meetings for backtesting.
How can equity and event-driven desks use prediction market data?
Kalshi and Polymarket price earnings beats, revenue thresholds, M&A closes, IPO timing and regulatory rulings on listed companies. Dune's ticker mapping links each of those markets to its ticker, so a desk can see what the market expects before a print and join it straight to its book. Ahead of NVIDIA's August earnings, traders priced a beat at 95.8%.
How can a commodities desk use prediction market data?
Kalshi lists daily contracts on where WTI will settle at each price level, and Polymarket prices supply shocks such as a Strait of Hormuz closure. Together they give a market-implied read on the events behind crude and gold moves. Every contract resolves against the actual market outcome, so implied and realized sit side by side for backtesting.
How can quant and risk teams use prediction market data?
Quant teams use hourly probability series and resolved outcomes back to 2021 to build and backtest event-driven signals, with no venue-specific cleaning. Risk teams screen holdings for event exposure that options may not price yet: every ticker-mapped market carries an event type and a confidence score.
How can desks track AI trade exposure with prediction market data?
Prediction markets price the AI trade from several angles: hourly GPU rental rates for H100, H200, B200 and RTX 5090 on Kalshi, which company has the top AI model each month, model release dates, whether NVIDIA or Alphabet ends each month as the world's largest company, and AI IPOs such as Anthropic. GPU rental contracts give a market read on compute demand, a direct input for semiconductor and hyperscaler exposure. Dune's ticker mapping files these markets under the names they concern: 1,858 markets mapped to NVDA in the last 90 days alone.