Historical price series, executed trades and recorded order books answer different questions. Prices show a quoted or observed level over time; trades show transactions; book states show the resting prices and quantities recorded by a capture system. Choose the input for your research task before choosing a provider.
ZenHodl sells archives, so this is a commercial buyer's guide. Free sources can be the right choice. A paid archive is useful only if its specific window, fields and limitations fit the question you want to answer.
Start with official data and free archives
The Polymarket market-data documentation separates price history from current books. Kalshi documents historical trades and candles, with separate cutoffs. A transaction or price series alone cannot reconstruct the unobserved resting ladder. Check the exact current endpoint contract instead of assuming all historical data is unavailable.
You also have access to third-party recordings:
| Source | Documented option | What to verify |
|---|---|---|
| PMXT archive | Free archived order-book data for Polymarket and Kalshi, among other venues | File dates, schema, timestamps, retained depth and gaps. Hourly file publication does not by itself establish an hourly observation interval. |
| Marketlens public repository | Free Polymarket books, trades and outcomes for selected events, including sports | Whether your particular contracts and research period are present; units, license and capture method. |
| Your own recorder | Books and messages observed after you start collecting | Sequence handling, clocks, reconnect gaps, storage, licensing and market selection. |
| A commercial recording provider | A bounded file release or a continuing data service | A sample and an actual coverage report for the product, not just its advertised row count. |
Those free sources were checked on October 8, 2026. Their availability does not establish complete coverage of any venue or season. It does mean that “record it yourself or buy it” is an incomplete description of the options.
Choose the right ZenHodl product
Our products have separate schemas and captured windows. Do not combine their features into one imaginary archive:
- The legacy Polymarket and Kalshi archive contains independent venue tables. It is not a matched panel and does not provide per-level ladders. Capture-time game context belongs to its Kalshi table, not its Polymarket table.
- The recent Polymarket sports archive contains top quotes and sampled L2 states. Its quote clock is recorder time, top-size validity is limited, books can repeat, and it contains no trade prints. Review the dated schema.
- The Kalshi depth and trade tape contains sampled historical depth, capped at 20 levels per side, and captured prints. Trade
created_timeis exchange time;capture_tsis retrieval time. Context attached to a retrieved print belongs to retrieval time. See the dated schema. - The MLB matched-book product is a separate game-keyed cross-venue panel. Its nearest captured Kalshi match can fall before or after the Polymarket observation. It is not proof of simultaneous executable quotes, and a causal study must exclude future observations. The Polymarket complement is derived, not independently captured; unresolved outcomes remain null. See its matching schema.
Use the current catalog for prices, packaged vintages and listed bounds. A release cutoff is not evidence that every market appears throughout the window. An unknown or unverified book bound should remain unknown until the delivered payload or trusted manifest establishes it.
Five checks before paying
- Research window. Ask for first/last observed book times and per-market presence. Separate exchange trade times, write partitions and archive inclusion cutoffs.
- Observation unit. Is a row a quote, a book level, a message, a trade or a game? Repeated states are not independent market events.
- Clock and alignment. Identify the exchange clock, recorder clock, quote age and join direction. Capture-time scores are not an event-time sports timeline.
- Depth and quantities. Check retained levels and known truncation. A displayed ladder can constrain a simulation; it does not establish queue position or fills.
- License and delivery. Read research, commercial-use and redistribution terms, and check how downloaded files are verified.
For a score join, a separately recorded and time-valid sports timeline can be joined afterward. Whether that join is possible depends on the identifiers, clock quality and historical timeline available; it is not categorically impossible, and it is not automatically accurate.
Inspect a sample first
Start with the free samples and capture-gap guide. The Kalshi tryout walkthrough demonstrates a dated reproducible audit. The Polymarket backtesting guide separates recorded observations from simulated execution.
Other vendors named in earlier versions of this guide include Telonex, PolymarketData, and Kalshi BackTest. Check their current primary documentation rather than treating the old comparison's prices or exclusions as current. No exhaustive vendor or feature audit is claimed here.
Updated October 8, 2026: corrected free-source availability, product distinctions and clock/alignment claims. This article does not report a new paid-payload study or promise a trading result.