Sample evidence memo · 2026-10-16
Hyperliquidity Provider (HLP)
vault 0xdfc24b077bc1425ad1dea75bcb6f8158e10df303 · data from api.hyperliquid.xyz, fetched 2026-10-16 10:18:45 UTC · raw numbers
The question
A depositor is looking at HLP because it is the strongest record on Hyperliquid’s all-time PnL board. How much of that record counts as evidence?
Short answerPromising, not proven
Strong for a record picked blind: t = 3.43 over 179 weeks. Not yet strong enough for the best of 47,310 accounts (bar 4.90); at its own Sharpe of 1.91 it needs about 38 more months. And what it shows is not a trader’s skill: it is the exchange’s own vault, which earns from market making, liquidations and platform fees.
1 · What the record is
- A Hyperliquid vault, described by Hyperliquid as: “This community-owned vault provides liquidity to Hyperliquid through multiple market making strategies, performs liquidations, and accrues platform fees.” Deposits are open. The leader holds 0.20% and takes 0% of depositors’ profits.
- It is the parent of 7 child vaults. 4 of them are also on the all-time PnL top 100 (#22 HLP Liquidator, #53 HLP Strategy A, #72 HLP Strategy B, #87 HLP Liquidator 2), so that board shows one operation 5 times.
- History 2023-05-10 to 2026-10-16: 98 scored intervals covering 179.2 weeks. Weeks with no new sample in Hyperliquid’s history are merged into the next interval rather than counted as flat.
- Coverage: HLP trades on HyperCore, so its whole PnL is in the record Hyperliquid keeps; the HyperEVM gap noted on the wallet page does not apply. Deposits and withdrawals change the account value, never the PnL.
2 · The numbers
| Measure | Value | Reading |
|---|---|---|
| t (Student’s t, 93 effective degrees of freedom) | 3.43 | Annualised Sharpe 1.91 × √3.4 years, adjusted for record length |
| Bar for a record picked blind | 2.00 | clears |
| Bar for the best of 47,310, accounts treated as independent | 4.90 | does not clear; about 38 more months at this Sharpe |
| Same bar, rough allowance for accounts that trade alike | ≈ 2.6 | clears; an order of magnitude, not a measurement |
3 · What t cannot see
- Shape of the record. 89% of intervals were positive. The worst three were -4.4%, -4.2%, -4.0%; the deepest drawdown, compounding the intervals, was 5.8%. Skew is +4.9: the largest moves were gains. That is not the shape of a strategy that wins small and loses big, at least so far.
- The tail that has not happened. A vault that takes over liquidated positions is on the other side when prices gap. The worst case it is exposed to need not appear in 3.4 years of history, and t cannot price it.
- Where the returns come from. Fees and liquidation flow are income from the venue’s own activity. For a depositor the question is less whether someone is skilled than whether that income continues, and on what terms.
4 · What would change the verdict
- More record: at a steady Sharpe of 1.91, about 38 more months to clear the best-of-47,310 bar.
- The forward record: HLP is one of the 200 wallets frozen on 2026-10-14. Its score on the record after the freeze is published daily from 2026-12-09.
- A drawdown well beyond the 5.8% seen so far, which would show the tail described above.
5 · Method and reproducibility
Same estimator as the wallet check; definitions in the method. To recompute: python tools/memo_sample.py 0xdfc24b077bc1425ad1dea75bcb6f8158e10df303 out.json --board-n 47310 from the repository.
This page shows the format of a one-page evidence memo, on a public record. It is not investment advice. HLP was not consulted and paid nothing; we never take money from the party being scored.