HyperliquidOrdersEventsL4 order-book data from HyperCore: every order status including cancels and rejects, per-order book deltas, and the oracle and mark price pushes behind them.
Hyperliquid order book data from the HyperCore L1, not HyperEVM. The order_updates table keeps every order through its lifecycle: open, canceled, filled, triggered and the *Rejected states, with time in force, reduce-only, trigger conditions and position TP/SL flags. The book_updates table is L4, one row per new, update or remove of a single order with its owner, order id, price and size, so the book can be rebuilt at any block rather than at snapshot boundaries. The price_updates table records each mark, oracle and external perp price per coin, and the inputs behind them.
Hyperliquid's own S3 archive publishes L2 snapshots and asset contexts and nothing else. Rejects and cancels are invisible in any trades-only dataset, and they are most of the signal in microstructure work. Money and sizes are decimal strings, addresses are 0x-prefixed hex, and Order_PlacedTime is epoch milliseconds while Block_Time and Order_EventTime are epoch nanoseconds.
One row per order status change: open, canceled, filled, triggered and every *Rejected state, with time in force, reduce-only and trigger fields. Order_EventTime is epoch nanoseconds.
hyperliquid/<start_date>_<end_date>/order_updates<start_block>-<end_block>-<random suffix>.parquet; file size variesL4 order-book deltas: one row per new, update or remove of a single order, with owner, order id, price and size.
hyperliquid/<start_date>_<end_date>/book_updates<start_block>-<end_block>-<random suffix>.parquet; file size variesOne row per coin and price kind in each oracle push: the resulting mark, oracle and external perp prices, and the inputs behind them.
hyperliquid/<start_date>_<end_date>/price_updates<start_block>-<end_block>-<random suffix>.parquet; file size variesorder_updates51 columnsOne row per order status change: open, canceled, filled, triggered and every *Rejected state, with time in force, reduce-only and trigger fields. Order_EventTime is epoch nanoseconds.book_updates36 columnsL4 order-book deltas: one row per new, update or remove of a single order, with owner, order id, price and size.price_updates20 columnsOne row per coin and price kind in each oracle push: the resulting mark, oracle and external perp prices, and the inputs behind them.Send this dataset’s full schema to an assistant and ask it anything. It reads the plain-text brief first, so the answer comes from the real column list rather than a guess.
Opens a new chat with the question filled in. Nothing about you is sent to Bitquery, and the assistant sees only the public brief.
Parquet files in a Requester Pays S3 bucket or a Cloudflare R2 bucket, chosen at checkout. On S3 you download with your own AWS account and AWS bills the transfer; on R2 you download with a read-only API key and pay no egress. Layout:
# order_updates hyperliquid/<start_date>_<end_date>/order_updates<start_block>-<end_block>-<random suffix>.parquet; file size varies # book_updates hyperliquid/<start_date>_<end_date>/book_updates<start_block>-<end_block>-<random suffix>.parquet; file size varies # price_updates hyperliquid/<start_date>_<end_date>/price_updates<start_block>-<end_block>-<random suffix>.parquet; file size varies
Three Parquet tables from HyperCore: order_updates, book_updates and price_updates, with 107 documented columns. Every row carries Block_Number and Block_Time, so the tables join on block with each other and with the Hyperliquid Trades and Funding datasets.
Large, and driven by quoting activity. In a measured hour of HyperCore on 29 September 2026 (50,417 blocks) we counted 76.6 million order statuses and 56.0 million book diffs, about 1.5 million and 1.1 million per 1,000 blocks, plus a few dozen oracle updates per 1,000 blocks. Scaled to a day that is roughly 1.8 billion order statuses and 1.3 billion book diffs. Volume moves with market activity; the file count and total size come with the delivery, and these figures are estimates.
Hyperliquid's archive holds L2 book snapshots and asset contexts, uploaded roughly monthly, and states that no other historical data is provided that way. This dataset is L4: each book_updates row names the order and the address behind it, and every open, cancel and reject is in order_updates.
Replay book_updates in Block_Number order per BookUpdate_Market_CoinRaw: Kind new adds an order at BookUpdate_Px with BookUpdate_Size, update changes its size from BookUpdate_SizeBefore to BookUpdate_Size, remove deletes it. Aggregate by price for an L2 view at any block, or keep order ids for queue position.
In the Hyperliquid Trades dataset. A filled status here tells you an order completed; the match price, fee and realized PnL are in its trades table, and liquidation fills in perp_liquidations, joined on Block_Number and the order id.
30 days. HyperCore history is kept as a rolling 30-day window, so a purchase covers the 30 days up to the day before we prepare the files, and there is no longer window to buy. For a continuous feed beyond that, ask about the live stream.
Because the node emits them that way and we keep the values as emitted. Block_Time and Order_EventTime are epoch nanoseconds; Order_PlacedTime is epoch milliseconds. Divide by 1e9 or 1e3 respectively when converting to timestamps.
With DuckDB: SELECT * FROM read_parquet('order_updates/*.parquet'). With pandas: pd.read_parquet('order_updates/'). Parse Order_LimitPx, Order_Size, BookUpdate_Px and the other price and size columns with Decimal rather than float; they are stored as decimal strings on purpose.
Yes. These field definitions are the ones Bitquery's Hyperliquid Kafka topics use, so a backtest reading this archive and a production consumer reading the live stream decode with the same code.
As Parquet files, one folder per delivery window and one prefix per table, for example hyperliquid/2026-08-31_2026-09-29/order_updates/. File names carry the block range they cover. We send the file count and total size with the delivery. We prepare the files within 7 business days of purchase, covering the 30 days up to the day before. You choose at checkout where to collect them: a Requester Pays S3 bucket, downloaded with your own AWS account, or a Cloudflare R2 bucket with a read-only API key and no egress fees. Files stay available for 7 days from delivery.
Yes. Every table links to a public sample Parquet file with the first 100 rows of a real slice and every column the export writes. The sample repository on GitHub shows the first records of each table as readable JSON. No email address required.
PolymarketFull history · dailyPolymarket fills, position splits, merges and payouts, and market results on Polygon since September 2025, with market titles, outcomes and USD values.
RobinhoodFull history
Every Pons launch, bonding-curve trade and graduation on Robinhood Chain, decoded from the chain and priced in USD.
ArbitrumFull historyArbitrum One since its genesis block of May 2021 as flat Parquet: blocks, transactions, internal calls, decoded events, transfers, DEX trades and sequencer fee rewards, at block resolution.
Tell us the chain, tables and date range.