HyperliquidEventsOrdersOracle price, mark price and hourly funding rate history for BTC, ETH, SOL, HYPE, XRP, DOGE and every other validator-operated HyperCore perp, block by block.
Native HyperCore perps are the markets the Hyperliquid validator set operates itself: BTC, ETH, SOL, HYPE, XRP, DOGE, ZEC, XMR, WLD, RUNE, TRUMP, LINK, NEAR, kPEPE and about 220 others (Market_Protocol = hypercore), as opposed to HIP-3 markets deployed by third parties. Their oracle is not published as a HyperCore event stream; it lives in the signed consensus actions that go into each block. This dataset parses those actions. native_price_updates is the SetGlobalAction the engine applies roughly every 40 to 50 blocks: one row per native market per push, with the oracle price and the mark price the engine set, the external perp price, and the block, consensus round and proposer. native_oracle_votes is each validator's validatorL1UpdateReferenceOracle submission: one row per validator, market and vote, so the aggregated oracle can be checked against its inputs. native_perp_fundings is the hourly settlement: one row per account and native market with the rate, signed position and USDC paid or received.
Every row carries Block_Number and Block_Time (epoch nanoseconds, UTC). Price rows also carry Bundle_Index and Action_Index, the position of the action inside the block, so ordering is deterministic down to the action. Prices are decimal strings exactly as the node recorded them; addresses are 0x-prefixed hex. HIP-3 markets are not in this dataset; their oracle and mark prices are in Hyperliquid Order Book.
One row per native perp market per SetGlobalAction push: the oracle and mark price the engine set, with block, round, proposer and action position.
hyperliquid/native_price_updates<start_block>_<end_block>.parquet1000 blocks per fileOne row per validator, native market and validatorL1UpdateReferenceOracle submission: the price that validator voted, with block, round and action position.
hyperliquid/native_oracle_votes<start_block>_<end_block>.parquet1000 blocks per fileOne row per account and native perp market at each hourly funding settlement: rate, signed position and USDC paid or received.
hyperliquid/native_perp_fundings<start_block>_<end_block>.parquet1000 blocks per file; one file per settlement hournative_price_updates19 columnsOne row per native perp market per SetGlobalAction push: the oracle and mark price the engine set, with block, round, proposer and action position.native_oracle_votes17 columnsOne row per validator, native market and validatorL1UpdateReferenceOracle submission: the price that validator voted, with block, round and action position.native_perp_fundings14 columnsOne row per account and native perp market at each hourly funding settlement: rate, signed position and USDC paid or received.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:
# native_price_updates: 1000 blocks per file hyperliquid/native_price_updates<start_block>_<end_block>.parquet # native_oracle_votes: 1000 blocks per file hyperliquid/native_oracle_votes<start_block>_<end_block>.parquet # native_perp_fundings: 1000 blocks per file; one file per settlement hour hyperliquid/native_perp_fundings<start_block>_<end_block>.parquet
Three Parquet tables from HyperCore consensus data: native_price_updates (19 columns), native_oracle_votes (17 columns) and native_perp_fundings (14 columns). All native perp markets, every block in the window. Each row carries Block_Number, so the tables join with each other and with the Hyperliquid Trades and Order Book datasets.
Every validator-operated HyperCore perp, about 235 markets in November 2026. By fills on 5 November 2026 the busiest were RUNE, ZEC, HYPE, WLD, BTC, LIT, ETH, SOL, TRUMP, INIT, XMR, NEAR, kPEPE, ICP, INJ and XRP; by notional BTC and ETH lead. Delisted markets such as MATIC carry no price and are omitted. HIP-3 markets (xyz:, io:, para: and the other deployer prefixes) are not included.
HyperCore settles funding once an hour at the first block after the hour. The oracle price in force at that block is the last native_price_updates row for that market with Block_Number at or below the settlement Block_Number in native_perp_fundings. Join on Market_Coin = Funding_Market_CoinRaw. The validator votes behind that push are in native_oracle_votes at the same blocks.
From the signed actions in the consensus stream (replica_cmds), not from the event streams, API snapshots or inferred from payments. SetGlobalAction is the engine's own oracle and mark price write; validatorL1UpdateReferenceOracle is each validator's submitted vote. The raw action is parsed verbatim; Consensus_Round, Consensus_Proposer, Bundle_Index and Action_Index say exactly where in the block it sat.
Our raw block archive has short holes. For the 30 days to 5 November 2026 they are: about 70 minutes on 8 to 9 October (blocks 1176454000 to 1176512999), and single 1,000-block holes (about 70 seconds each) on 7, 10 and 25 October plus five consecutive ones on 14 October. Blocks inside a hole are absent from all three tables; a hole never lands on a funding settlement block in that window. Gaps cannot be recovered after the fact, so the delivery manifest lists them explicitly.
SetGlobalAction lists prices by position in the HyperCore perp universe (0 = BTC, 1 = ETH, 2 = ATOM and so on). We resolve the position to Market_Coin using the universe as published by the node, and keep Asset_Index so the mapping can be checked. Delisted positions carry no price and are omitted.
For a measured 1,000-block pack on 5 November 2026: 3,938 price rows (22 pushes), 72,336 vote rows (411 votes) and, at a settlement block the same morning, 310,226 funding rows across 179 native markets. Scaled to 30 days that is roughly 150 million price rows, 2.7 billion vote rows and 220 million funding rows, around 12 GB of ZSTD Parquet in total. These are estimates; exact counts and sizes come with the delivery manifest.
30 days up to the day before we prepare the files. HyperCore history is kept as a rolling window, and there is no longer window to buy. For a continuous feed beyond that, ask about the live stream.
With DuckDB: SELECT * FROM read_parquet('native_price_updates/*.parquet'). With pandas: pd.read_parquet('native_price_updates/'). Parse Oracle_Px, Mark_Px, Funding_Amount, Funding_Rate and Funding_Size with Decimal rather than float; they are stored as decimal strings on purpose.
As flat Parquet files, one folder per table, one file per 1,000-block pack, with a JSON manifest listing every file, its sha256 and the block holes. We prepare the files within 7 business days of purchase. 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.
HIP-3 markets (their oracle and mark prices are in Hyperliquid Order Book), fills, order and book data, spot and outcome markets, and ledger events.
PolymarketFull history · dailyPolymarket fills, position splits, merges and payouts, and market results on Polygon since September 2025, with market titles, outcomes and USD values.
SolanaLast 12 months
Every Pump.fun token launched since October 2025, with its creator, name, symbol and URI at launch, every graduation to PumpSwap, and per-creator launch records.
RobinhoodFull history
Every Pons launch, bonding-curve trade and graduation on Robinhood Chain, decoded from the chain and priced in USD.
Tell us the chain, tables and date range.