# Hyperliquid Builder Fees | Bitquery Data Store Daily Hyperliquid volume and fees per builder code and market, with HIP-3 deployer fees, for perps, HIP-3 stock perps and spot over the last 30 days. Dataset page: https://bitquery.io/datastore/datasets/hyperliquid-builder-fees Network: Hyperliquid Category: Trades, RWA Tables: 1 Columns: 15 Coverage: first block to 2026-11-15 Rows: about 380K Size: about 19 MB Format: Apache Parquet, ZSTD compression, one prefix per table Licence: https://bitquery.io/datastore/legal/data-license ## What this is Hyperliquid builder codes tag every order an app sends, and the exchange pays that app a builder fee on each fill. This table sums HyperCore fills per UTC day, builder code, market and fee token: fills, distinct traders, volume, and the fee split between the builder, the HIP-3 deployer and the exchange. Fills sent without a builder code are in the same table with an empty Builder_Address, so shares add up to the whole exchange. Most buyers sum Volume and Builder_Fees by Builder_Address over a week, or filter Market_Kind = 'hip3' to rank stock perps by volume and deployer fees. On 11 November 2026 builder-coded fills were $1.40 billion of $20.36 billion in fill volume and paid $423,000 in builder fees. Builder codes are addresses; we do not attach app names. ## What people use it for - Rank apps by Hyperliquid volume routed through their builder codes - Builder fee revenue per app, market and day - HIP-3 deployer fees and volume per stock, index and commodity perp - Share of exchange volume that comes through front ends and bots ## Tables (1) ### hyperliquid_builder_fees_daily: 15 columns One row per UTC day, builder code, market and fee token: fills, traders, volume and the fee split between builder, HIP-3 deployer and exchange. Bucket prefix: hyperliquid/hyperliquid-builder-fees/hyperliquid_builder_fees_daily File naming: .parquet, one file per UTC day Free sample (real Parquet, no email needed): https://bitquery-blockchain-dataset.s3.us-east-1.amazonaws.com/hyperliquid/hyperliquid-builder-fees/hyperliquid_builder_fees_daily/sample.parquet Sample updated: Nov 16, 2026 | column | type | description | | --- | --- | --- | | Block_Date | Date | UTC day of the fills | | Builder_Address | String | Builder code address the orders were sent with, 0x-prefixed hex; empty for fills sent without a builder code | | Market_CoinRaw | String | Market as the node names it: plain perp (HYPE), HIP-3 perp (xyz:NVDA) or spot pair (@107) | | Market_Symbol | String | Readable market name: HYPE, or NVDA for xyz:NVDA | | Market_Kind | String | perp, hip3 or spot | | Market_Protocol | String | hypercore for plain perps and spot; the deploying dex (xyz, io, para, mkts) for HIP-3 markets | | Fee_Token | String | Token the fees in this row are paid in: USDC on perps, the received token on spot | | Fills | UInt64 | Fill sides; a match between two accounts is two fills | | Taker_Fills | UInt64 | Fill sides that took liquidity | | Traders | UInt64 | Distinct accounts behind the fills | | Volume | Float64 | Sum of price times size over the fills, in the quote currency (USDC on perps) | | Total_Fees | Float64 | Fees paid on the fills, in Fee_Token; includes the builder fee and the HIP-3 deployer fee; maker rebates are negative | | Builder_Fees | Float64 | Part of Total_Fees paid to the builder code, in Fee_Token | | Deployer_Fees | Float64 | Part of Total_Fees paid to the HIP-3 market deployer, in Fee_Token; 0 on plain perps and spot | | Exchange_Fees | Float64 | Total_Fees minus Builder_Fees and Deployer_Fees: what the exchange kept, net of maker rebates | ## Price - Latest month (Oct 15, 2026 → Nov 15, 2026): $150 one-time, USD One-time purchase. The files are yours to keep under the licence above. ## Questions buyers ask Q: What is included? A: One table, hyperliquid_builder_fees_daily, with 15 columns. Each row is one UTC day, builder code, market and fee token, with fills, taker fills, traders, volume and four fee columns. Plain perps, spot and HIP-3 perps from the xyz, io, para and mkts dexes are covered. Liquidation fills are left out. Q: How big is it? A: About 12,000 to 13,000 rows a day, so around 380,000 rows for 30 days. One day is about 600 KB of ZSTD Parquet, so the whole window is around 18 MB. These are estimates from our 14 November 2026 sample; the delivery manifest gives exact figures. Q: Which apps route the most Hyperliquid volume? A: Filter Builder_Address <> '' and sum Volume and Builder_Fees grouped by Builder_Address. On 11 November 2026, 536 builder codes had fills. The table gives addresses only, so match them to apps from the builders' own announcements. Q: Which stocks trade on Hyperliquid? A: Filter Market_Kind = 'hip3' and group by Market_Protocol and Market_Symbol; the xyz dex lists most stock perps, as xyz:NVDA, xyz:TSLA and so on. Sum Volume for activity and Deployer_Fees for what the deployer earned. Q: How are the fees split? A: Total_Fees is the fee on the fill and already includes the builder fee and, on HIP-3 markets, the deployer fee. On an ETH taker fill we checked, the fee was 8.67 basis points of notional, of which 5 went to the builder. On an xyz:NVDA taker fill it was 5.86 basis points: 5 to the builder, 0.39 to the deployer, the rest to the exchange. Q: How are the numbers checked? A: The Fills total for a day equals the number of fill rows in our HyperCore trades table for that day: 13,148,003 on 11 November and 13,255,502 on 12 November 2026. Q: What limits should I know about? A: Volume counts each side of a match, so the sum over all rows is twice the matched volume. Fees on spot are in the received token, so sum fees within one Fee_Token. HIP-3 dexes margined in USDH, USDE or USDT0 (km, flx, vntl, hyna, cash) are not in our HyperCore feed yet. Q: How do I load it? A: With DuckDB: SELECT Builder_Address, sum(Volume), sum(Builder_Fees) FROM read_parquet('hyperliquid_builder_fees_daily/*.parquet') WHERE Builder_Address <> '' GROUP BY 1 ORDER BY 2 DESC. With pandas: pd.read_parquet('hyperliquid_builder_fees_daily/'). Q: How far back does it go and how is it delivered? A: 30 days, the rolling window our HyperCore history keeps, up to the day before we prepare the files. Files are Parquet, one per UTC day, ZSTD-compressed, in a Requester Pays S3 bucket; download costs are billed to your AWS account. ## Data after your purchase A purchase is a one-time set of files, up to the coverage end date on the dataset page. For data after that: - Real-time stream: Kafka Streams sends new Hyperliquid data as it happens. Enterprise plan. Topics and schema: https://docs.bitquery.io/docs/streams/kafka-streaming-concepts/#hyperliquid-hyperliquid - Daily warehouse tables: Datashares refreshes tables daily in Snowflake, BigQuery, S3 or Azure. Enterprise plan. https://bitquery.io/products/datashares - Another one-off slice (other chains, tables or dates): a custom extract. https://bitquery.io/products/data-on-demand ## What this file is, and is not This is a description of one dataset sold by Bitquery, written for assistants and for people. It lists every column but holds no data rows. The sample files linked under each table are real Parquet and are free to download. Figures here come from the product record and are exact unless marked otherwise; the delivery manifest is authoritative for a purchased file. If a question needs a row that is not in a sample, say so rather than guessing at it.