The Hyperliquid HLP vault — what it does, where its money comes from, and what you are actually buying
Most people meet HLP as a number on a screen: a vault with a percentage next to it, sitting one click away from the trading interface. That framing is the problem. HLP is not a savings account with a rate; it is a live trading book that quotes both sides of the market and takes over positions when someone else’s liquidation fails. Depositing means owning a slice of that book. This guide takes the venue’s own documentation apart to show where each dollar of HLP’s profit comes from, which parts of it you can verify and which you cannot, and what the four-day lock is really protecting.
Key takeaways
- HLP is a trading book, not a yield product. Depositors own a proportional share of a market maker that also inherits failed liquidations. The documentation is explicit that you earn a share of the profits or losses, and that past performance guarantees nothing.
- Its liquidation business lives inside a very narrow price window. Ordinary liquidations run through the public book; HLP only steps in below two thirds of the maintenance margin. On a BTC position that is a band 0.41% of price wide — $415 on a $100,000 entry.
- No profit share is a bigger edge than it sounds. User vaults take 10% of profits; protocol vaults take nothing. Compounded over twelve periods at the same gross return that gap is 20.6% of the total gain, not 10%.
- The four-day lock restarts on every deposit. Not on your first — on your most recent. Adding to the position every four days or less means never being able to withdraw at all.
What does HLP actually do with the money in it?
Four jobs, and the Protocol vaults page lists them in one sentence: HLP “provides liquidity to Hyperliquid through multiple market making strategies, performs liquidations, supplies USDC in Earn, and accrues a portion of trading fees.” It then makes the ownership claim that explains why the vault exists at all — HLP “democratizes strategies typically reserved for privileged parties on other exchanges” and is “fully community-owned”.
Read that list again as a risk statement rather than a feature list, because that is what it is.
- Market making means standing on both sides of the book and holding whatever inventory the market hands you. A market maker in a calm range collects the spread; a market maker in a one-way move ends up long into a fall or short into a rally, because that is what taking the other side means.
- Performing liquidations means acquiring positions from traders whose margin has run out. Every one of those positions was, moments before, losing money fast.
- Supplying USDC in Earn puts idle collateral to work — a mechanism the venue names here but does not describe on a page of its own, which is worth noting rather than guessing about.
- Accruing a portion of trading fees is the only genuinely passive line, and it is also the one you cannot size. More on that in a moment.
Nothing in that description resembles a deposit. Two of the four jobs are directional risk taken on purpose, and depositors hold them in proportion. The depositor page puts the arrangement in a single line: “By depositing, you earn a share of the profits, or losses, of the vault.” The comma before “or losses” is doing a lot of work, and it is the most important punctuation on the page.
Where does HLP’s money come from, and how much of it can you verify?
Three of the four income lines can be reasoned about from public rules. One cannot, and pretending otherwise is how most write-ups of HLP go wrong.
Fees: documented, unquantifiable. The Fees page says that on Hyperliquid “fees are entirely directed to the community (HLP, the assistance fund, and deployers)”, in contrast to venues where “the team or insiders are the main beneficiaries”. What it never publishes is the split. HLP accrues a portion, and the size of that portion appears nowhere in the documentation. So the honest statement is: a base-rate market-order round trip on $10,000 costs $9.00 at 0.045% each way, all of it goes to the community, and no reader can work out how much of it reaches HLP. Any article that gives you a percentage there is filling a gap with invention.
Market making: visible, not forecastable. The venue runs no designated market maker programme — its Market making page states there is “no DMM program, special rebates / fees, or latency advantages” — so HLP competes for maker rebates on the same published schedule as everyone else. At the top of that schedule, maker volume above 3% of the venue’s total earns a rebate of 0.003%. That is real money at size and it is also thin enough that a single bad inventory day erases weeks of it.
Liquidations: the part with actual arithmetic. This is where HLP’s design is genuinely unusual, and it is the next section.
Earn: named but undocumented. The vault supplies USDC into it. There is no dedicated page describing terms, counterparty or rate. We are flagging the gap rather than filling it.
One comparison makes the fee point concrete. On the three large centralised venues we track, perpetual taker fees sit at 0.05% (0.055% on Bybit) against maker 0.02%, all checked against their public schedules in the first half of September 2026. Hyperliquid’s base perpetual taker rate is 0.045% and maker 0.015%. The rates are in the same neighbourhood. The difference that matters is not the number but the destination: on a centralised venue that fee is revenue for a company, and on this one the documentation routes it to a vault you are allowed to own part of.
How does HLP make money from a liquidation, and how narrow is that window?
Narrower than almost anyone realises, and the arithmetic is the most useful thing on this page.
Start with what does not reach HLP. When account equity falls below the maintenance margin, the Liquidations page says market orders for the full position are sent to the public order book, where anyone may fill them, and if the requirement is met again the remaining collateral stays with the trader. That is the normal case, and it is also why the venue can say there is no clearance fee on liquidations.
HLP only appears at the next stage. “If the account equity drops below 2/3 of the maintenance margin without successful liquidation through the book, a backstop liquidation happens through the liquidator vault” — and that vault is “a component strategy of HLP”. The position and its margin transfer across, and the remaining equity is not returned. The documentation explains why without softening it: the vault “requires a buffer to make sure backstop liquidations are profitable on average.”
Now put the two triggers on a price scale. Take a $10,000 BTC long at 40x isolated, entered at $100,000 for clean arithmetic. Maintenance margin on BTC is 1.25% of the position, so the liquidation trigger is the price at which equity falls to $125 and the backstop trigger is the price at which it falls to two thirds of that.
The gap between them is 0.41% of price — $415 on a $100,000 entry. That is the entire distance in which the public order book has to absorb a failing position before HLP inherits it. Extend the same calculation across every maintenance band the venue publishes and a pattern appears that nobody puts in a marketing page.
| Maximum leverage on the asset | Maintenance margin | On a $10,000 position | Backstop trigger | Width of the window | Assets in that band |
|---|---|---|---|---|---|
| 40x | 1.25% | $125 | $83.33 | 0.415% | BTC |
| 25x | 2.00% | $200 | $133 | 0.662% | ETH |
| 20x | 2.50% | $250 | $167 | 0.826% | SOL, XRP |
| 10x | 5.00% | $500 | $333 | 1.633% | most mid-caps |
| 5x | 10.00% | $1,000 | $667 | 3.175% | above the tier limit |
| 3x | 16.67% | $1,667 | $1,111 | 5.000% | the least liquid assets |
The window is narrowest on the deepest market and widest on the thinnest one. On BTC the book gets 0.41%; on the least liquid assets, capped at 3x, it gets a full 5% — twelve times as much room. That is the correct design and it is worth saying so plainly: a deep book needs less room to clear a position than a thin one, so the buffer scales with how hard the job is.
It is also the source of the risk depositors are taking, because the design assumes price moves through the window rather than jumping over it. A BTC move of 0.41% is not an event; it is a bad minute. When price gaps that distance faster than the book can fill — and the venue’s own partial-liquidation rule adds a 30-second cooldown after any partial liquidation above $100,000 of notional — HLP inherits a position with less cushion than the model assumes, at exactly the moment its market-making inventory is also on the wrong side. The two jobs fail together, or not at all. That correlation is the central fact about HLP’s risk, and it does not appear in any single number on the vault page.
Why does HLP charge no profit share when user vaults charge 10%?
Because it is not a business, structurally. The HyperCore vaults page sets both rules in consecutive sentences: “Vault owners receive 10% of the total profits”, and then “protocol vaults do not have any fees or profit share.” There is no owner to pay, because the vault is the protocol.
The documentation works its own example for a user vault: deposit 100 USDC into a vault holding 900, watch it grow to 2,000, withdraw 200 less a 10 USDC profit share for a net 190. Sound small? It is — once. A share taken every period is a different animal.
After one period the 10% share costs exactly what it says, $20.00. After twelve it has taken $1,629 — 20.6% of the entire gain, because every dollar it removes is also a dollar that stops compounding. This is the standard fee-drag argument applied to a structure most readers have not seen it applied to, and it is the strongest single reason to prefer the protocol vault over a user vault with a similar published record.
It is not, however, a reason to prefer HLP over holding USDC. A zero fee on a negative return is still a negative return. The comparison the fee structure wins is against other vaults, not against not depositing.
What does the four-day lock-up actually cost you?
More than four days, because of one clause most summaries drop. The lock runs from your most recent deposit, not your first. The documentation spells it out with a date: deposit on Sep 14 at 08:00 and you can withdraw on Sep 18 at 08:00. User-run vaults lock for one day; HLP locks for four.
Now apply that to the way people actually fund a position — in instalments.
| If you add to the deposit every… | Days in each cycle you could withdraw | Share of the time you can get out | What that means |
|---|---|---|---|
| 2 days | 0 | 0% | permanently locked |
| 4 days | 0 | 0% | permanently locked — the clock restarts as it expires |
| 7 days | 3 | 43% | locked on four days out of seven |
| 14 days | 10 | 71% | the lock is a minor constraint |
| 30 days | 26 | 87% | the lock barely binds |
Adding money every four days or less means never being able to withdraw, because each deposit restarts the clock on the whole balance rather than on the new money. Nothing in the documentation flags this as a trap and it is not one — it follows from one sentence — but it converts a mild-sounding constraint into a permanent one for anyone dollar-cost-averaging weekly or faster.
The more interesting question is what the lock is for. HLP is the venue’s backstop liquidator. A backstop that can be emptied by depositors at the first sign of a cascade is not a backstop. Four days is, roughly, the length of a serious deleveraging event in crypto — and it is the window in which HLP would be inheriting positions fastest. Read structurally, the lock exists precisely so that depositors cannot leave during the event they would most want to leave during. That is a defensible design for the protocol and a cost the depositor is paying, and both halves of that sentence are true at once.
One more reason the lock matters: withdrawing is not free of market impact either. The depositor page notes that “there may be some slippage as you withdraw and open positions are closed”. A large exit from a vault holding inventory closes that inventory into the market.
How should you judge HLP before depositing anything?
Four checks, in order, all of which you can do from the public vaults page without connecting anything.
- Read the drawdown before the APY. The vault page publishes profit and loss, maximum drawdown, volume, open positions and trade history. The headline percentage describes a past the vault cannot repeat on request; the drawdown describes a loss the same strategy already produced once. If the drawdown figure would have made you withdraw, and the lock would have stopped you, you have your answer already.
- Look at the open positions, not just the curve. This is the part a centralised insurance fund never shows you. A vault whose open positions are mostly one-way inventory in a trending market is telling you where its next loss comes from.
- Size it as a trading allocation, not as cash. You cannot get out for four days and you share losses in full. Money that has a job inside that window does not belong here.
- Decide whether you want the correlation. If you also trade perpetuals on the venue, HLP is not a diversifier — it is a second position that loses money in the same conditions as your first. A liquidation cascade hits your margin and HLP’s inventory simultaneously.
And one structural point that belongs with those four. The auto-deleveraging page closes with a guarantee: “a user who has no open positions will not socialize any losses of the platform.” That is a genuinely strong protection, and it is written for accounts that are flat. HLP, by construction, is never flat. It sits in the auto-deleveraging queue like any other account holding positions, which means a profitable HLP position can be closed at the previous mark price against an underwater trader. Depositors are buying the one account on the venue that can never qualify for its best protection. That is not a flaw in the design — somebody has to be there — but it should be priced, and it usually is not.
IF YOU WANT TO READ THE VAULT PAGE FOR YOURSELF
Everything above is checkable before you commit a dollar: the drawdown, the open positions, the trade history and the deposit lock are all on the public vaults page, and the rules behind them are in the documentation linked throughout. Two things are not on that page, and they are the arrangement you are accepting. You hold the keys yourself — a lost seed phrase is lost money, with no support desk to appeal to and no company that can reverse a mistake. And if you trade here as well as deposit, the venue checks your leverage once, when the position opens, and never again; the form starts at 20x cross on BTC, which is a 3.80% move from liquidation. Open the vaults page, read the drawdown first, and decide the size before you decide anything else.
Referral link — the venue pays us a share of its trading fees at no extra cost to you, and it pays nothing on vault deposits, so this link earns us nothing if you only deposit. It does not change a number on this page. Education only; most retail traders lose money.
Where can this cost you?
Where it can cost you: the structure is more transparent than any insurance fund on a centralised exchange, and three things about it still work against a depositor. All three are in the documentation. None of them is in the summary anyone reads first.
- You cannot decompose the return. The venue publishes that HLP accrues a portion of trading fees and never publishes the portion. So when the vault is up, no depositor can say how much came from fee accrual, how much from market making and how much from liquidations — which means no depositor can tell whether the strategy is working or whether fee income is covering trading losses. That is a real gap in an otherwise unusually open set of documents.
- The lock is longest exactly when you would want it shortest. Four days, restarting on every top-up, in a vehicle whose job is to absorb liquidation cascades. The design reason is sound. The cost lands on you.
- “Community-owned” describes ownership, not control. Depositors share the profit and loss. They do not set the strategies, the inventory limits or the fee split, and the documentation describes no mechanism by which they could. It is an open cap table on a book somebody else runs.
Anyone who found HLP by looking for a yield on idle stablecoins. The word “vault” and a percentage next to it produce a mental model — deposit, accrue, withdraw — that this product does not fit at any point. If the sentence “I am buying a share of a market maker’s book” does not feel accurate to you, the next step is what a perpetual actually is, not a deposit.
Anyone who cannot leave the money alone for four days, or who would add to it more often than that. The lock is not negotiable and it restarts every time you top up.
Anyone who has not yet read how liquidation works on this venue from the trader’s side. HLP’s income is the other half of that mechanism, and the two pages only make sense together: start with funding and liquidation on Hyperliquid, then come back.
Common mistakes
- Treating the displayed APY as a rate. It is a backward-looking figure produced by a trading strategy, not a yield anyone has promised. The documentation says outright that past performance is not a guarantee of future returns.
- Assuming the four-day clock starts at the first deposit. It starts at the most recent one, on the whole balance. Weekly top-ups leave you locked four days out of every seven.
- Thinking HLP earns the full maintenance margin on every liquidation. Most liquidations never reach it — they clear on the public book and the trader keeps what survives. HLP only takes the ones that fail below two thirds of the maintenance margin.
- Reading “no profit share” as “low risk”. The fee structure is excellent and it says nothing whatever about the risk of the strategy underneath it.
- Using HLP as a hedge for your own perpetual positions. It is the opposite of a hedge. A cascade damages your margin and the vault’s inventory in the same hour.
- Depositing without opening the drawdown chart. The single most predictive number on the page is the one that describes the largest loss the strategy has already taken, and the lock means you would have had to sit through it.
Frequently asked questions
What is HLP on Hyperliquid?
HLP, the Hyperliquidity Provider, is a protocol vault. The documentation describes it as providing liquidity through multiple market making strategies, performing liquidations, supplying USDC in Earn, and accruing a portion of trading fees. It is community-owned: anyone can deposit USDC and take a proportional share of its profit and loss, and unlike a user-run vault it charges no fees and no profit share.
Is depositing into HLP safe?
It is not a deposit product and the documentation does not present it as one. You are buying a share of a live trading book that market-makes and inherits liquidated positions, so you share its losses exactly as you share its profits. The venue’s own line is that trading is inherently risky and past performance is not a guarantee of future returns. On top of market risk, USDC sits in a smart-contract system you do not control and cannot withdraw from for four days after your most recent deposit.
How long is money locked in HLP?
Four days from your most recent deposit, not from your first. The documentation gives the example directly: deposit on Sep 14 at 08:00 and you can withdraw on Sep 18 at 08:00. User-run vaults lock for one day instead. The practical consequence is that topping up restarts the clock on the whole balance, so adding money every four days or less keeps you permanently locked.
Does HLP charge a fee or a profit share?
No. The HyperCore vaults page states that vault owners receive 10% of total profits and then adds that protocol vaults do not have any fees or profit share. Applied to the documentation’s own worked example — a 100 USDC deposit in a vault that doubles — the difference is a 200 USDC withdrawal instead of 190. Compounded across twelve periods at the same gross return, a 10% share absorbs 20.6% of the gain.
How does HLP make money from liquidations?
Most liquidations never reach it: they are sent to the public order book as ordinary market orders that anyone may fill, and whatever collateral survives stays with the trader. Only when account equity falls below two thirds of the maintenance margin does the liquidator vault — a strategy inside HLP — take the position over, and the remaining equity is not returned. The documentation says plainly that the vault needs that buffer so backstop liquidations are profitable on average.
Can HLP lose money?
Yes, and the structure says how. It holds inventory as a market maker and it inherits positions that were losing badly enough to be backstopped, so a fast one-way move is exactly the condition in which both jobs go wrong at once. The auto-deleveraging page also guarantees that a user with no open positions never socialises platform losses — which is precisely the protection a vault holding positions around the clock can never qualify for.