Hyperliquid frontends builder revenue leaderboard

The USD value of builder fees each Hyperliquid frontend collected on the last complete UTC day, and over the 7 and 30 days ending on it. A builder fee is the optional rebate the frontend writes into the builder field on every Hyperliquid order it routes. The on-chain protocol caps this at 100 basis points on perps and 1000 on spot. The bench sums every fill's builderFee value attributed to a known frontend address and publishes the total per timeframe.

The previous version of this bench ranked frontends by effective fee in basis points, which answered the trader question of cost per dollar routed. The dollar revenue framing answers the operator question of size and revenue scale, the same way CoinMarketMan and ASXN dashboards present builder leaderboards. Both views are useful. The effective fee bps view is still available as a companion tab.

From the per-builder daily fills feed Hyperliquid publishes on its public stats CDN: one CSV per builder address and UTC day, with every fill that carried that builder code. A Go harness on OCB infrastructure downloads each file once the day is published and aggregates it locally. No node, no API key, no third party indexer, no internal Mobula service. Source code is in `harnesses/hyperliquid-frontends` in this repo.

The harness keeps the last 30 daily files per builder in a local mirror. The 7 day figure sums builder fees over the files of the 7 UTC days ending on the current feed day, the 30 day figure over 30 days. Wallet counts for those windows are unions of the daily wallet sets. Files older than the window are deleted; a small ledger of daily totals is kept for the revenue history on the product pages.

The feed is cut per UTC day, so the shortest window available is one complete day: midnight to midnight UTC. The 24h label is kept for continuity with the metric names and the recorded history. Every builder is measured on the same day, which is fairer than a rolling window that zeroes frontends whose flow lands early in the day.

Every Hyperliquid frontend with a known builder address, 104 today. The cohort spans consumer wallets (Phantom Perps, MetaMask, Rabby, OneKey, Trust Wallet), pro terminals (Insilico, Axiom, Pear, Lit Trade), social and pvp UIs (pvp.trade, Senpi, FOMO), bots and signal apps (FlowBot, Tread fi, Moonbot, Hyperdash), based.app and defi.app. The registry is `harnesses/hyperliquid-frontends/builders.json` in this repo. New addresses ship via public PR.

Three reasons. The frontend stopped operating or routed no attributed fills that day. The frontend never enabled the builder code attribution system. The frontend uses a builder address not yet in our registry. A frontend at zero on the daily view but not on 7 day or 30 day is in the first bucket; a consistent zero across all timeframes points to the second or third.

Consumer wallets integrate Hyperliquid trading directly inside the wallet UI, so their user base is large; pro terminals typically charge less per dollar routed by design, even when their notional volume is comparable. The Effective fee bps tab shows each frontend's actual rate, and the leader is recomputed from the live figures on every render.

The harness checks the feed every 30 minutes and Prometheus scrapes the gauges every 30 seconds, but the numbers only change once a day, when Hyperliquid publishes the previous UTC day (early morning UTC). The timestamp at the top of the page is the end of the day the numbers describe.

Yes. Every number is a Prometheus query exposed via the OCB API endpoints. The query string is visible in the row hover tooltip on the leaderboard. The harness source is open at the link in the source field below. Cite the value and the timestamp at the top of the page.