Prediction markets produce something researchers usually have to beg for: a complete, timestamped record of what a crowd believed, priced in money, with the outcome attached. The polymarket api and the kalshi api both expose most of that record for free. If you study forecasting, market microstructure, or how news moves beliefs, this is the most accessible new dataset in finance, and almost nobody in academia has fully mined it yet.
Polymarket's three APIs
Polymarket splits its data across three services, and knowing which does what saves hours. The Gamma API is the public catalog: market questions, outcomes, current prices, volume and end dates, readable with no account and no key. The CLOB API is the exchange itself: live order books, a price history endpoint for charting and backtests, and the trading endpoints, which do need credentials. The Data API serves trades, positions and leaderboard history.
| API | What it serves | Access |
|---|---|---|
| Gamma API | Market metadata, outcome prices, volume, tags | Public, no key |
| CLOB API | Order books, price history, trading | Reads public; orders need credentials |
| Data API | Trade history, positions, holder data | Public |
The kalshi api and its free history
Kalshi's REST API mirrors its market structure: series contain events, events contain markets, and each market carries a ticker. Market data is public without a key, including order books, candlestick history and the public trade tape; placing orders requires signed authentication with an API key pair. The part researchers underuse is the historical layer: archived markets, historical trades and minute-level candlesticks reach back months, which makes a kalshi historical data download a realistic weekend project rather than an enterprise negotiation. WebSocket feeds stream live updates if you would rather record the market yourself.
Ready-made datasets, dashboards and research use
You do not have to start from the raw APIs. Services such as DeltaBase and Kingsets republish recent Polymarket and Kalshi trades as clean CSV files and public BigQuery datasets, free for recent windows, with paid tiers for tick-level history and full order book archives. Dune hosts community dashboards built on Polymarket's on-chain settlement data, since every trade on the global platform settles on Polygon in the open. Academic snapshots have started appearing too, including a 200-market dataset on IEEE DataPort built from the Gamma API and price history endpoints, with resolutions attached for calibration studies.
The research questions practically queue up: Are market prices calibrated, and where do they systematically overreact? How fast does news enter prices compared with social media? Do the culture and AI markets forecast better than expert panels? Prediction market data for research has one giant advantage over survey data: the respondents have money down.
Getting started takes an afternoon: list open markets from the Gamma API or Kalshi's markets endpoint, pick a handful with real volume, fetch their price history, and join it to the recorded resolution. Both APIs speak plain JSON over HTTP, community client libraries exist for Python, and rate limits on the public endpoints are workable if you cache what you fetch and poll politely.
Free vs gated: what you actually need
For most studies, the free tier is the whole job. Live prices, market metadata, resolutions and sampled price history cover calibration, accuracy and event studies. You pay, or record your own feed, only when you need tick-by-tick order book history, which neither exchange gives away complete. One warning applies to both platforms: record each market's rules and resolution source alongside its prices. Two markets with the same headline can settle differently, and that detail, not the price series, is what ruins datasets. It is the same care this site takes with its own numbers, from the earnings calendar to the country rankings.
Kalshi is a CFTC-regulated exchange and Polymarket's US access is rolling out state by state, but the data described here is public reading material on both. Pull an API response, and you will understand these markets faster than a week of commentary could teach you. Or if you would rather see finished analysis than raw feeds, the Global Portfolio Builder is this site's own small example of market data turned into a tool.