Running a public Hyperliquid vault since February: the numbers
The short answer: RockSteady, our strategy, has run as a public vault on Hyperliquid since 3 February 2026. In its first 9 months it grew about 10% after trading costs, before the 10% fee on gains, leaving out deposits and withdrawals. Its deepest dip was 9.9%, and it took 203 days to get back to a new high. 4 of its 9 months were down. Every figure here comes from the vault itself, so you can check it.
We publish this because a vault’s history is the only honest answer to “does it work?”. Backtests can be tuned until they look good. A live vault cannot be edited afterwards.
The vault in 1 table
| RockSteady | |
|---|---|
| Name on Hyperliquid | Agent RockSteady |
| Vault address | 0xd321ad79041bdad719ba5ff42fe950e1977df439 |
| Opened | 3 February 2026 |
| Growth since opening | about +10% |
| Deepest dip so far | 9.9%, from 4 March to 8 April 2026 |
| Back to a new high | 23 September 2026, 203 days after the high before the dip |
| Trades closed | 230 by 30 October 2026 |
| Fee | 10% of gains, set by Hyperliquid |
| Lockup | 1 day after each deposit |
| Our own money in the vault | about 9% of it |
Figures as of 31 October 2026. The live record updates them from the vault every day.
What RockSteady does
RockSteady trades crypto perpetual futures on Hyperliquid by fixed rules, around the clock. It mostly buys sharp drops in price and sells into the rebound, with some short trades, spread across many coins. It aims for many small winning trades rather than a few big ones, and usually holds a trade for hours.
The rules went through more than 30 versions on 5 years of 1 minute price data, 2021 to 2025, and were frozen before launch. Nothing has been tuned since the vault opened. AI tools helped with research and testing; live trading follows the fixed rules and nothing else.
Month by month
| Month | Growth |
|---|---|
| February (from the 3rd) | +4.0% |
| March | +0.7% |
| April | −4.2% |
| May | +2.9% |
| June | +2.0% |
| July | −0.4% |
| August | −0.5% |
| September | +7.2% |
| October | −1.4% |
| All 9 months | about +10.2% |
Hyperliquid records a vault about once a week, so each month here ends at its last weekly reading, not on the last day. Growth is chained reading by reading, so money moving in or out never counts as a gain or a loss. The months multiply to the total; they do not add up to it.
The pattern is what a dip buying strategy should look like: a few strong months when sharp drops snap back, flat or slightly negative months when prices drift. Most of the year’s gain came in 2 months, February and September.
The worst stretch
The vault hit a high on 4 March, then fell 9.9% by 8 April as crypto prices kept sliding instead of bouncing. A strategy that buys drops does worst in exactly that kind of market: it buys, the price keeps going, and it takes the loss.
It then took until 23 September, 203 days from the March high, to reach a new high. That is more than half the vault’s life so far spent below its best level.
We did not change the rules during that stretch. Changing a strategy in the middle of its worst run is how a tested strategy turns into a guess. In testing, the deepest dip over 5 years was 23.7%, so 9.9% was inside what we expected. That does not mean the next dip stops at 9.9%.
If you are thinking about any vault, ask its leader the same 2 questions: how deep has it fallen, and how long did it take to recover? A vault that shows only its return is hiding half the picture.
What we got wrong
The description on the vault page on Hyperliquid was written at launch. It says the strategy was picked from 800 tested agents and is powered by Claude. Hyperliquid lets a vault’s description be written once, when the vault is created, and never changed, so we cannot correct it.
What is true today: the strategy was built and tested with AI tools, then frozen, and simple fixed rules trade it live. No AI makes live trading decisions. Our check page says the same, next to the vault’s details.
Check every number yourself
You do not need to trust this post. 3 places show the same vault, and we control only one of them:
- Hyperliquid: the vault page shows every trade, every deposit and withdrawal, the fee and the leader’s share.
- Trading Strategy, an independent vault data site: RockSteady on Trading Strategy shows its returns, deepest dip and fee from its own copy of Hyperliquid’s data.
- Our live record, read from the same vault.
Small differences between them come from the start date and the method. Hyperliquid’s own chart starts at the first reading, on 4 February.
Is about 10% in 9 months good for a Hyperliquid vault?
It depends on what it cost to get there. About 10% with a deepest dip of 9.9% is a modest return for a modest dip. Many Hyperliquid vaults show bigger returns; check how deep they fell to get them. Compare return and deepest dip together, never return alone. Our guide to Hyperliquid vaults explains how to read a vault page.
Can I withdraw at any time?
Yes, 1 day after each deposit. You can withdraw in the NuWealth app, or directly on Hyperliquid without our app at all. You get your share of what the vault is worth that day, minus 10% of any gain.
Can anyone see the trades?
Yes. Every vault on Hyperliquid is public. Anyone can see RockSteady’s open positions, past trades, deposits and withdrawals on its vault page, with no account needed.
What is next
RockSteady keeps running on the same rules, and the live record keeps updating, good months and bad.
Past results are not a promise of future results. A strategy that follows fixed rules can lose, sometimes for months. Read our Risk Disclaimer before you deposit.