KnownAt
Sign inGet API key
ProductPolymarketEarly accessPredict.funIn developmentHyperliquidIn developmentMacro vintagesIn developmentAll datasets
ResourcesDocumentationBlogChangelogPricing

Alternative data infrastructure

Live and historical data
for bots and backtests.

Query prediction markets, on-chain activity, and macro releases through one API—cleaned, timestamped, and ready for research.

Get API key

Free workspace · No card required

DATASET PREVIEWpolymarket.market_snapshots
POINT-IN-TIME EARLY ACCESS
Select a view
observed_attimestampmarketstryesfloat64volume_usdfloat64
12:00:000x3a67…cc140.4181,284,402
12:01:000x3a67…cc140.4201,286,119
12:02:000x3a67…cc140.4231,291,805
12:03:000x3a67…cc140.4251,294,240
12:04:000x3a67…cc140.4271,301,552
GET/v1/datasets/polymarket.market_snapshots/latestJSON  CSV  PARQUET
03launch datasets
LIVE + HISTORYone query model
POINT-IN-TIMEno silent overwrites
API + FILESJSON · CSV · Parquet

01 / THE DATA LAYER

Research the market.
Not the data pipeline.

KnownAt turns fragmented sources into consistent datasets you can query live, download in bulk, and replay historically.

01

Stop maintaining scrapers

We collect, normalize, validate, and monitor each source continuously.

02

Backtest what existed then

Original observations and later revisions remain separately addressable.

03

Keep one query model

Move from current observations to historical replay without rebuilding your integration.

02 / LAUNCH CATALOG

Choose the market.
Keep the interface.

Each source gets its own documented dataset, coverage policy, schema, and quality metrics. The access layer stays the same.

01 / Prediction marketsSources: Polymarket, Predict
Early access

polymarket.market_snapshots

Market prices, liquidity, volume, lifecycle events, and resolution history.

Source history
2020 → now
Update
1 min target
Live APIHistoricalPoint-in-time
Polymarket Predict.fun
02 / On-chain marketsSources: Hyperliquid
In development

hyperliquid.l2_book

Order book updates, trades, funding, and market state in one normalized schema.

Source history
2023 → now
Update
1 sec target
StreamingHistorical replayL2
Explore Hyperliquid
03 / Macro vintagesSources: FRED, BEA, BLS
In development

fred.macro_releases

Economic releases preserved exactly as they were originally published.

Source history
1959 → now
Update
Per release
RevisionsPublication timeAs-known-at
Explore macro vintages

03 / ONE INTERFACE

Change the dataset.
Not your workflow.

Use the same primitives for the latest observation, bounded history, and point-in-time replay.

Python
knownat
dataset = knownat.dataset(
  "polymarket.market_snapshots"
)

rows = dataset.history(
  from_="2024-11-01",
  to="2024-11-06",
  as_known_at="2024-11-06T00:00:00Z"
)
REST JSON CSV Parquet
query result
UTC · point-in-time
observed_attimestampavailable_attimestampsourcestrmarket_idstrvaluefloat64
2024-11-05 12:00:002024-11-05 12:00:01polymarket0x3a67…cc140.4180
2024-11-05 12:01:002024-11-05 12:01:01polymarket0x3a67…cc140.4200
2024-11-05 12:02:002024-11-05 12:02:01polymarket0x3a67…cc140.4230
2024-11-05 12:03:002024-11-05 12:03:01polymarket0x3a67…cc140.4250

04 / POINT-IN-TIME BY DEFAULT

Backtest what was known then.
Not what we know today.

KnownAt preserves when a record was observed, published, available, and ingested—plus every later correction.

Typical historical APIlatest value
History silently changes

Revisions overwrite the record your strategy would actually have seen.

observedrevisedoverwritten
KnownAtas_known_at=
Every vintage stays queryable

Replay the exact information set available to the model at any moment.

observedpublishedpreserved

05 / OPERATIONAL PROOF

Inspect the data
before you trust it.

Coverage, missing intervals, ingest delay, provenance, and revision policy belong in the product—not in a support ticket.

Per-record provenanceEvery row points back to its source.source_record_id
Append-only historyCorrections never erase the original.revision_policy: append_only
Visible data healthFreshness and gaps are queryable.GET /v1/quality
Production metrics will be published per dataset as ingestion pipelines enter service.Read the build log

06 / PRICING

One plan while we build
the launch catalog.

Founding access covers the first production dataset, full history, and API access. No sales call and no feature maze.

Founding

Early access

Full access to the first KnownAt dataset while the catalog grows.

···/ monthCancel at any time. No trial period.
  • One production dataset at launch
  • Daily data updates
  • Historical downloads
  • Point-in-time API access
  • JSON, CSV, and Parquet

Prices exclude tax. Local VAT or sales tax is calculated and added at checkout.

07 / BUILD LOG

What changed,
and when.

Read research notes
v0.3.0

Platform catalog architectureOrganized the product around launch datasets and a shared access layer.

v0.2.0

Live-to-history data modelDefined capture, provenance, revision, and as-known-at query semantics.

append-only

Corrections never erase the originalAffected records and replacement vintages remain addressable.

BUILD ON THE DATA

Start querying
alternative markets.

Get API key