Data
Hyperliquid's most underrated feature is not its latency. It is that the entire trading history of every account is queryable, which turns questions that are normally matters of trust into matters of computation.
In short
Because HyperCore holds the order books, margin state and clearinghouse onchain, every account's fills, open positions, funding payments and liquidations are public and queryable through Hyperliquid's API servers. That data is enough to reconstruct any trader's realized PnL series from raw fills, without their permission or cooperation — the basis of independent trader scoring.
For any address you can obtain the fill history, current open positions with entry and leverage, funding payments, and liquidation events, alongside market-level data such as the book, oracle and mark prices, and funding history.
There is no privileged tier of this data. The same state that the matching engine uses is the state anyone can read, which is the direct consequence of not running an off-chain order book.
A displayed PnL number is not a track record. The useful sequence is: pull the account's fills, group them into positions, close positions into realized trades, and build a timestamped realized PnL series. Only then can you ask whether the curve is broadly distributed or a single outlier with noise around it.
Unrealised PnL is deliberately excluded from that series. An open position is a claim, not a result — and an account can carry a large unrealised gain right up to the moment it is liquidated.
Hyperliquid serves this data through API servers with rate limits, and sub-account and vault activity is attributed distinctly, which matters when you are attributing volume or performance. Any production system reading it needs caching and backfill rather than polling naively.
Refer to Hyperliquid's own API documentation for current endpoints, limits and schemas; those change more often than the underlying guarantees do.
Public fills show what an account did, not why. They cannot reveal off-exchange hedges, coordinated accounts run by the same desk, or a change in the strategy behind an unchanged address. A regime shift can invalidate a strong record with no visible failure at all.
This is why scores are recomputed continuously and why the sticky basket has explicit soft-issue and emergency replacement rules, rather than treating a score as a durable label.
Methodology
Scoring and replacement are documented in full on How it works and in the Docs (Policy v3). In short: the Elite basket is sticky, emergencies remove a leader immediately, and soft issues accrue at most one strike per UTC day with three strike-days triggering replacement. Read how it works or the documentation for the full table.
Questions
Yes. Orders, cancels, fills and liquidations are onchain state on HyperCore, attributable to the account that produced them.
Yes — pull their fills, reconstruct closed positions, and build a realized PnL series. That is exactly what independent scoring does, and it needs no cooperation from the trader.
Intent, off-exchange hedges, and coordination between addresses. It also cannot tell you whether the strategy behind an address has quietly changed.
Score-weighted allocation across up to 10 elite Hyperliquid traders, each isolated in its own sub-account. Your funds never leave your account.
Non-custodial · Agent cannot withdraw · Cancel delegation anytime
