Export Format
Data Catalog exports are delivered as Apache Parquet with ZSTD compression. Date ranges use UTC dates. Schema availability depends on venue family, market, coverage, and plan or purchase path. The cart matrix below covers the current selectable file schemas; individual market pages can expose a narrower set based on coverage.Schema Matrix
Candles and order flow have no standalone export schema. For files, export
trades and aggregate to candles, or export l4_orders and aggregate to order flow.
Schema availability is per venue: l2_orderbook and trades exist on every venue; l3_orderbook is Lighter-only; l4_orderbook and l4_orders are Hyperliquid-family only; liquidations is core and HIP-3 only; funding is core, HIP-3, and Lighter; oi is core, HIP-3, HIP-4, and Lighter. The cart shows the selectable set for each market.
Lighter funding records are exportable where the cart offers that schema, but currently served values must not be compared across venues or annualized pending normalization repair.
Export Schema Checklist
Use this checklist before mapping an export to a warehouse table or model input.Coverage And Range Rules
Catalog schema availability starts from the symbols feed row. A schema is selectable when the row’sdata_types includes the schema key, the schema coverage key, or a documented alias. l2_orderbook maps to orderbook, and oi can map to open-interest coverage when an upstream surface uses that name.
For each selected schema, use coverage_by_type[schema.coverageKey] when it exists, then fall back to coverage_from. When multiple schemas or HIP-4 outcome sides are selected together, the valid export start date is the latest coverage start across the selected rows and schemas. For settled HIP-4 outcomes, cap the export end date at coverage_to; for live markets, use today as the default end date.
The Data Catalog defaults to a 90-day range when coverage allows it, otherwise it uses all available history. Custom ranges still have to stay inside the selected schema coverage window.
Hyperliquid core trades and liquidations, plus Hyperliquid Spot trades, should not be requested before 2025-03-22 through the export flow even when a broader market row has older coverage metadata.
How Schemas Map To API Concepts
Order-book exports can represent different depths. L2 snapshots are not L3 individual orders or L4 per-order book records. L4 order-book and separate L4 order-event exports should stay distinct from ordinary order-book JSON responses in code, warehouses, and docs. Trades, liquidations, funding, and open interest can have different coverage starts for the same market. HIP-4 exports need side and event context. Outcome prices should be treated as probability-like values, not ordinary USD prices. HIP-3 exports need the builder prefix preserved. Spot exports need pair symbols. Lighter exports should stay under the Lighter venue family.Choosing Schemas
1
Start with the research question
If the question only needs top-of-book or daily aggregates, do not default to L4.
2
Check coverage by schema
Schema history can differ within the same market.
3
Estimate size before checkout
High-depth schemas can be much larger than funding or OI datasets.
4
Store schema keys with output
A downstream warehouse table should know the schema key and venue family that produced it.