Polymarket and Kalshi historical data API: where to get it
Where to get Polymarket and Kalshi price history: what each API returns, its time steps, its limits and cost, plus one tool for both.
Both venues publish price history through their own public APIs. Polymarket's prices-history endpoint needs no key, but one start-to-end request covers at most 15 days, and fine steps only reach back 30 days. Kalshi's candlestick endpoint gives bid, ask, trade price, volume and open interest in 1-minute, 1-hour or 1-day steps. Comparand's venue_prices tool returns either venue's price history in one format, as a probability from 0 to 1.
This guide compares what each one says it offers. These are the tools' own published rules, not a speed or accuracy test.
What does a prediction market price history show?
A prediction market lets people trade on whether something will happen. A contract pays out if the event happens and nothing if it does not. Its price runs between 0 and 1.
A price 62 percent of the way from 0 to 1 reads as the market's rough chance that the event happens: 62 out of 100. A price history is that number over time. A candle (also called a bar) is the open, high, low and close price over a set time, such as one hour.
Price history is useful for research. You can see how the market's view moved after news, how fast it moved, and what the price was at a given hour. It is a record of what traders paid, not proof of what was true.
How do the three sources compare?
| Kalshi API | Polymarket APIs | Comparand venue_prices | |
|---|---|---|---|
| What history it returns | Candles: open, high, low, close for yes bid and ask, plus trade price, volume, open interest | Price points for each outcome token | A price series for one market, as a probability from 0 to 1 |
| Time steps | 1, 60 or 1,440 minutes | Buckets of 60 to 86,400 seconds | 1 minute, 1 hour (default) or 1 day |
| Covers | Kalshi only | Polymarket only | Kalshi and Polymarket |
| Key and cost as published | No subscription price on the pages we read; rate-limit tiers instead | No API subscription price seen; public market data needs no credentials | Free tool; needs a Comparand key |
| Main published limit | Older data sits behind separate historical endpoints | Range limited to 15 days; fine buckets only for the last 30 days | Fair-use request limits on your plan |
Sources: Kalshi candlesticks, Kalshi historical data, Polymarket prices and order books and the Comparand docs.
What does Kalshi's API do well?
Kalshi gives the most detail per candle. Each candle has open, high, low and close for the yes bid and ask, plus the trade price, volume and open interest. Open interest is the total size of trades still open. If you care about the spread between bid and ask, Kalshi's own data shows it.
Kalshi's docs list REST, WebSocket and FIX access, and the index also lists trades, order book, event candlestick and batch candlestick endpoints. That is more than a price series.
Two limits to know about:
- Old data moves. Older settled markets, candlesticks and trades sit behind separate historical endpoints. The cutoff date advances over time, so you read it from a cutoff endpoint.
- Some numbers are not stated. The pages we read do not give a history start date or a maximum candle count per request.
Rate limits depend on your tier. The read budget per second runs from 200 tokens at Basic to 12,000 at Prestige, and most requests cost 10 tokens. Going over the budget returns a 429 error.
What does Polymarket's API do well?
Polymarket splits its data into three public REST APIs. Gamma finds events and markets. CLOB serves prices and order books. Data serves positions, activity and participation. Public market data needs no credentials.
The prices-history endpoint returns price points for each outcome token. You can choose a recent interval, an exact start and end, or a single moment. Bucket width runs from 60 to 86,400 seconds. A full-history request comes back in 12-hour buckets by default.
The limits are the part to plan around:
- A start-to-end range is limited to 15 days.
- Finer buckets return only the last 30 days.
- Polymarket states it keeps at least 7 days of 1-minute data, 60 days of 5-minute data and 90 days of 30-minute data. Its 3-hour and 12-hour data is permanent.
So how do you get a long Polymarket history? Going by these rules, you have two choices. Ask for the full history and take the 12-hour buckets. Or ask for 15 days at a time at a finer step, but only inside the last 30 days. A long history at a fine step is not something this endpoint is built for.
Rate limits are per IP address over 10-second windows. Examples listed are 200 requests for one prices-history route, 1,000 for another and 300 for Gamma markets. Both venues also document live feeds (Polymarket's real-time streams, Kalshi's WebSocket feeds). Those are for live prices, not history.
Where does Comparand's venue_prices fit?
venue_prices returns the price history of one Kalshi or Polymarket market as a probability from 0 to 1. You give it the venue, the market, a start and an end. Steps are 1 minute, 1 hour (the default) or 1 day. It is free on every key.
Its main use is one format for both venues. venue_markets lists the markets on a venue, and you can pass a search word. venue_coverage shows which venues, datasets and steps are served now. All six venue tools, including these, are free, and they work through the API or inside AI apps like Claude, ChatGPT and Cursor through MCP.
What it does not do, as described in its docs: it returns a price series, so it does not list bid and ask, volume or open interest. Kalshi's candlesticks do that better. It does not state a start date for each market here, so check venue_coverage and test your market. A Comparand key comes with a plan or a first pay-as-you-go purchase (from $5), and plans carry fair-use request limits of 60 per minute on Starter, 300 on Pro and 900 on Desk.
For a wider comparison of crypto data sources, see crypto market data APIs compared.
Which one fits which job?
- You only trade or study Kalshi and need bid, ask, volume or open interest: use Kalshi's own API.
- You only study Polymarket and need recent fine steps or outcome-level prices: use Polymarket's own API, within its 15-day and 30-day limits.
- You want one format for both venues, or want to ask an AI app in plain words: use venue_prices. See how to connect Claude or ChatGPT.
- You need long history at a fine step for either venue: test one market first. The published limits differ, and no page we read promises it.
What does this comparison not cover?
We did not download price series from any of these tools. So we cannot tell you about gaps, speed or whether two sources agree on the same hour.
Community dashboards and third-party sites are not covered. We also do not quote trading fees, because this guide is about reading history, not placing trades. A prediction market price is a trader-set number, not a forecast of the outcome.
None of the official pages we read describes a search for past price charts that look like today's.
How we measured this
We read each tool's official documentation and recorded only what those pages say. Where a page does not state something, such as Kalshi's history start date, we say so rather than guess. Limits and prices change, so check the linked pages before you build on them.
Try this in Comparand
Ask your connected AI agent: "Which prediction markets does Comparand serve now? Then list Polymarket markets that mention my topic, and show one market's price for the last week in 1-hour steps."
- venue_coverage lists the venues and prediction markets served now, with their datasets and steps. It is free.
- venue_markets lists the markets one venue offers. You can pass a venue and a search word. It is free.
- venue_prices returns that market's price history as a probability from 0 to 1, with 1-minute, 1-hour or 1-day steps. It is free.
- Each answer says where the data came from and what the call cost.
You can also try 3 coins free to see how the past-chart search works for crypto. Setup is in the docs, and the pricing page lists the plans. These are past cases, not a forecast.
The short version
- Kalshi's API gives the most detail per candle: bid, ask, trade price, volume and open interest.
- Polymarket's API gives outcome prices, but caps a range at 15 days and fine steps at the last 30 days.
- Comparand's venue_prices gives both venues in one format, as a probability from 0 to 1, but only as a price series.
- Check each source's start date for your market before you rely on it.
- A prediction market price is a trader-set number, not a forecast.
Questions people ask
Where can I get Polymarket historical price data?
Polymarket's own prices-history endpoint returns price points for each outcome, in buckets from 60 to 86,400 seconds wide. Public market data needs no credentials. A start-to-end range is limited to 15 days, and the finest buckets only reach back 30 days.
Does Kalshi have a historical data API?
Yes. Kalshi's candlestick endpoint returns open, high, low and close for the yes bid and ask, plus trade price, volume and open interest, in 1-minute, 60-minute or 1,440-minute steps. Older settled markets, candlesticks and trades sit behind separate historical endpoints.
Can one tool give me both Kalshi and Polymarket prices?
Comparand's venue_prices tool returns the price history of one Kalshi or Polymarket market as a probability from 0 to 1, with 1-minute, 1-hour or 1-day steps. It is free on every key. It gives a price series only, so use Kalshi's own candlesticks if you need bid, ask, volume or open interest.
What does a prediction market price of 62 out of 100 mean?
On the 0 to 1 scale, that is a price 62 percent of the way up. Read it as the market's rough chance that the event happens. It is a price set by traders, not a measured fact, and it can move a lot as news arrives.