What One Key Returns
| Historical job | 0xArchive path | What you get |
|---|---|---|
| Hyperliquid order books, live and historical | /v1/hyperliquid/orderbook/BTC/history | 20-level snapshots about every 1.2 seconds since April 2023, 24.6B+ records; full-depth L2 since March 2026 |
| Hyperliquid trades, funding, OI | /v1/hyperliquid/trades/BTC, /v1/hyperliquid/funding/BTC, /v1/hyperliquid/openinterest/BTC | Trades since April 2023, continuous to the current hour; funding and OI since May 2023 |
| Order-level depth (L4) and lifecycle | /v1/hyperliquid/orderbook/BTC/l4/history, /v1/hyperliquid/orders/BTC/flow | Every add, fill, cancel, and TP/SL with wallet attribution since March 2026 |
| Liquidations | /v1/hyperliquid/liquidations/BTC | Since December 2025 |
| Lighter market data | /v1/lighter/trades/BTC, /v1/lighter/orderbook/BTC/history | Trades, funding, and OI since August 2025; native L2 tick since January 2026; L3 since March 2026 |
| Freshness, gaps, incidents | /v1/hyperliquid/freshness/BTC, /v1/data-quality/incidents | Coverage and freshness state to gate a window before a model trusts it |
Why Teams Choose 0xArchive
Backtests break when historical data cannot be tied to venue semantics, timestamp windows, route families, and data-quality state. 0xArchive keeps those concerns together so a research result is reproducible. A Hyperliquid backtest starts with/v1/hyperliquid/orderbook/BTC/history for 20-level book snapshots since April 2023, adds /v1/hyperliquid/trades/BTC for the matching tape, pulls /v1/hyperliquid/funding/BTC for carry, and gates the window with a /v1/hyperliquid/freshness/BTC check before the model trusts it. For order-level work, /v1/hyperliquid/orderbook/BTC/l4/history returns every add, fill, and cancel with wallet attribution since March 2026, the depth a DIY collector cannot rebuild after the fact.
Every response carries a request ID and a data-quality path you can check before a backtest trusts the window. Store route, symbol, parameters, cursor chain, request IDs, and freshness result with the dataset and the pull is repeatable months later. Every order, trade, and fill. One API.
Run One Route
Evaluation Questions
Choose the venue family
Start with Venue coverage. Hyperliquid core, Spot, HIP-3, HIP-4, and Lighter each keep their own history; do not normalize them into one symbol namespace.
Choose the time primitive
Use REST history routes for bounded pulls with explicit UTC start and end. Use WebSocket replay when event order and timing are the requirement.