Best Polymarket historical data APIs
Full matrix — official Polymarket APIs plus PolyOrderbooks, PolymarketData, Telonex, DepthFeed, Predexon, Dune, PolyBackTest, PolyTest, and polyReplay. Pricing, L2, Python, and bulk export.
Open provider matrixResearch
Start here
Before comparing features, it is worth checking whether a provider's data is internally consistent at all. The provider comparison has a four-line test you can run on any sample.
Full matrix — official Polymarket APIs plus PolyOrderbooks, PolymarketData, Telonex, DepthFeed, Predexon, Dune, PolyBackTest, PolyTest, and polyReplay. Pricing, L2, Python, and bulk export.
Open provider matrixPrefer describing the strategy to writing the replay loop? Backtest AI turns a plain-English idea into a run against the same L2 archive, with AI analysis of every backtest.
Alternatives
Migration guide after Dome’s end-of-life — what to replace for historical Polymarket workflows.
Read comparison →Side-by-side on L2 resolution, API limits, pricing model, and developer workflow.
Read comparison →Queryable REST archive vs daily Parquet tick pipelines — when each workflow fits.
Read comparison →Event-driven full-depth capture vs query-resolution historical APIs.
Read comparison →On-chain analytics vs stored L2 — why Dune SQL cannot rebuild the off-chain book.
Read comparison →BTC/ETH Up/Down full-depth order book backtesting vs ongoing 250ms Polymarket crypto L2.
Read comparison →Sub-second BTC/ETH/SOL Up/Down backtesting with 8-level depth vs broader 250ms full-ladder coverage.
Read comparison →Multi-venue data layer vs a Polymarket-specialized historical archive.
Read comparison →BTC/ETH Up/Down tick specialization vs broader Polymarket historical API coverage.
Read comparison →Weather market intelligence vs a historical crypto L2 archive — a category split with published pricing.
Read comparison →Rankings
Ranked APIs for live prices, 250ms historical L2, 1-minute breadth, and bulk.
Read ranking →Official API through enterprise archives — coverage, resolution, and pricing.
Read ranking →Archives ranked by job: 250ms L2, on-chain fills, 1-minute L2, tick Parquet.
Read ranking →Who captures depth, at what fidelity — 250ms on through tick-level Parquet.
Read ranking →Ranked by whether the price series ships with the book behind it.
Read ranking →Depth endpoints compared on history, depth, and grid — the matrix facts.
Read ranking →The live plane for execution and the historical plane for testing, split honestly.
Read ranking →Which archives preserve executable depth and which windows are too shallow.
Read ranking →Depth studies, microstructure, flow, and on-chain settlement — sized samples.
Read ranking →SDKs, MCP servers, SQL APIs, dashboards, and Parquet pipelines by workflow.
Read ranking →What each free tier really grants — a trial versus a working pipeline.
Read ranking →State from the official API, history-under-construction from archives.
Read ranking →Five families that survive honest backtesting on bound-sitting books.
Read ranking →What an MCP gives AI agents, and the honest limits of agent data access.
Read ranking →Regulated academic tool vs permissionless 250ms order-book venue.
Read comparison →CFTC-regulated event contracts vs a permissionless book — fees and data.
Read comparison →Play-money mana lab vs a USDC order book worth backtesting.
Read comparison →Forecast tournament with Brier scores vs a traded book.
Read comparison →Bonus-backed betting vs fee-free USDC order book.
Read comparison →Vig, liquidity, and cadence against a 250ms recorded book.
Read comparison →300ms BTC/ETH/SOL spec vs the wider 250ms archive, on the facts.
Read comparison →A 15-minute flat CSV dataset vs a queryable 250ms archive.
Read comparison →A backtesting-first service compared straight on the deciding factors.
Read comparison →It depends on workflow: official Polymarket APIs are free and best for price history and current books; archived full L2 needs a third-party provider. PolyOrderbooks, PolymarketData, DepthFeed, and Predexon serve REST historical L2; Telonex and polyReplay serve tick/Parquet archives. See the provider matrix for side-by-side facts.
The official API documents GET /book for current L2 order books and /prices-history for historical price series, but not a documented arbitrary historical full-L2 archive endpoint. Third-party providers store historical order books and expose them over REST, CSV, Parquet, or S3 delivery.
Provider facts are taken from each provider's current public documentation and pricing pages at a stated verification date, and are reverified periodically. Measured claims (such as replay accuracy) state their sample size and method on the page.
Yes — a CSV export of the same provider fields is available on the best historical data APIs page and the datasets area, with no signup required.
Prefer spreadsheets? Export the current provider matrix as CSV — same fields as the on-site tables, no signup required.