Historical Data

HyperLiquid provides historical data through S3 archive buckets and maintains several indexer tables for querying historical events.

S3 Archive Buckets

Historical data is available through AWS S3 buckets using requester-pays access:

Bucket	Content
`s3://hyperliquid-archive/market_data/`	L2 book snapshots and asset contexts. Updated monthly, LZ4 compressed.
`s3://hl-mainnet-node-data/node_fills_by_block`	Current fills stream organized by block
`s3://hl-mainnet-node-data/explorer_blocks`	Explorer block data
`s3://hl-mainnet-node-data/replica_cmds`	L1 transaction blocks (raw committed actions)

S3 access uses requester-pays, meaning the downloader pays for data transfer costs. The market_data bucket contains monthly updates with LZ4-compressed snapshots.

Data volume

The L1 produces approximately 100 GB per day of raw data across all output types. State snapshots are taken every 10,000 blocks.

Indexer Tables

The L1 indexer maintains several tables for historical event queries:

Table	Description
`UserFills`	Per-user trade fill history
`AssetTrades`	Per-asset trade history
`FundingHistory`	Historical funding rate payments
`AssetCtxHistory`	Historical asset context (OI, mark price, funding)
`OrderHistory`	Full order lifecycle history
`UserTwapSliceFills`	TWAP order slice execution history
`BoleInterest`	Borrow/lend interest accrual events
`LegacyRewardHistory`	Historical reward distributions
`ExplorerUserTxs`	Explorer-indexed user transactions

These tables are accessible through the info API endpoints (userFills, userFillsByTime, userFunding, historicalOrders, userTwapSliceFills, etc.)

State Snapshots

Full exchange state snapshots are serialized in MessagePack format. These snapshots capture the complete exchange state at a point in time, including:

All order books (perp and spot)
User positions and balances
Funding state
Bridge state
Governance/staking state
EVM state

Non-validator nodes can request state snapshots to bootstrap without replaying the entire history.

Data Formats

L1 Data Schemas

The L1 node produces several data types per block:

Output	Format	Description
`node_trades`	JSON	Executed trades with full side info
`node_fills`	JSON	Fill events with user attribution
`node_order_statuses`	JSON	Order lifecycle events (30+ status types)
`node_raw_book_diffs`	JSON	Order book mutations (new/update/remove)
`node_twap_statuses`	JSON	TWAP order progress updates
`misc_events`	JSON	Deposits, delegations, withdrawals, rewards, funding
`hip3_oracle_updates`	JSON	HIP-3 deployer oracle price updates

Trade format

Each trade record includes:

Trade:
  coin        — asset identifier
  side        — buy or sell
  time        — execution timestamp
  px          — execution price
  sz          — execution size
  hash        — transaction hash
  side_info   — per-side details (user, start_pos, oid, twap_id, cloid)

Order status format

Order lifecycle events include the full order struct plus a status field with 30+ possible values covering placement, fills, cancellations, and rejections.

Historical data volumes are substantial. The fills stream alone can produce gigabytes per day during active markets. Plan storage and processing capacity accordingly.

Signal Research

Protocol Research

DeFi Infrastructure

Infrastructure

S3 Archive Buckets

Data volume

Indexer Tables

State Snapshots

Data Formats

L1 Data Schemas

Trade format

Order status format

​S3 Archive Buckets

​Data volume

​Indexer Tables

​State Snapshots

​Data Formats

​L1 Data Schemas

​Trade format

​Order status format

S3 Archive Buckets

Data volume

Indexer Tables

State Snapshots

Data Formats

L1 Data Schemas

Trade format

Order status format