Hyperliquid's vault list has an APR column, and HLP, the exchange's own vault, sits at the top of it. The trouble is that the column looks like a savings rate and behaves nothing like one. When we checked in October 2026, the HLP vault showed an APR of roughly 3.5% on about $178 million of deposits, and a large share of the vault's lifetime profit arrived in a handful of violent days rather than as a steady drip.
That is not a reason to avoid it. It is a reason to understand what you are buying before you deposit. This review covers what Hyperliquid vaults are, where HLP's returns come from, how deposits and withdrawals work per Hyperliquid's own docs, the incidents that cost depositors money, and who the product genuinely suits. Every figure is anchored to October 2026, because every one of them moves.
What Hyperliquid vaults are
Hyperliquid is a perpetual futures exchange running on its own blockchain, where the order book, margin and liquidations are computed onchain by the network's validators rather than on a company server. A vault is a pooled trading account on that exchange. You deposit USDC, you own a proportional share of the pool, and your share rises or falls with whatever the vault's trading produces. In the words of the depositor docs, depositing means you "earn a share of the profits, or losses, of the vault."
There are two kinds. Protocol vaults are run by Hyperliquid itself, and HLP (Hyperliquidity Provider) is the one that matters. User vaults are run by individual traders or market makers, known as vault leaders, who trade the pooled money and keep a share of the profits. Hyperliquid's docs now call this original design legacy vaults and steer new builders toward a newer tokenized format. This review covers the original kind, meaning HLP and the trader-run user vaults listed in the app, because that is what most people mean by a Hyperliquid vault.
| HLP (protocol vault) | User vaults | |
|---|---|---|
| Who runs it | Hyperliquid's own strategies, community owned | An individual trader or market maker |
| What it does | Market making, backstop liquidations, supplying USDC to Hyperliquid's Earn lending, a share of trading fees | Whatever the leader trades, limited to validator-operated perps |
| Fee or profit share | None | 10% of each depositor's profit, paid to the leader |
| Lock-up after a deposit | 4 days | 1 day |
| Leader's own stake | Not applicable | At least 5% of the vault at all times |
Every row comes from Hyperliquid's own documentation, specifically the pages on protocol vaults, vault leaders and the legacy design, which states plainly that "protocol vaults do not have any fees or profit share."
Where HLP's returns come from
HLP is several strategies under one roof. Per the protocol vault docs, it "provides liquidity to Hyperliquid through multiple market making strategies, performs liquidations, supplies USDC in Earn, and accrues a portion of trading fees." Market making means quoting both sides of the order book and earning the spread between them, the same business we unpack in our guide to market makers. In calm markets that is most of what HLP does, and it pays modestly. It is a different engine from the funding and basis trades in our arbitrage explainer, because HLP is taking the other side of real order flow rather than locking in a price gap.
Liquidations are where the large numbers come from. A trader using leverage has to keep a minimum cushion of collateral behind each position, called the maintenance margin. When an account drops below it, Hyperliquid first tries to close the position by sending market orders to the order book, and most liquidations end there. Only if the account falls below two thirds of its maintenance margin without being closed on the book does the liquidator vault, a component strategy of HLP, step in as the backstop and take the position over, according to the liquidation docs. In that case the trader's maintenance margin is not returned and stays with the vault as a buffer. The docs say that buffer is what makes backstop liquidations profitable on average, and that on Hyperliquid the profit goes to HLP depositors rather than to the exchange operator.
The flip side is built into the same mechanism. HLP inherits positions at exactly the moments nobody else wants them. Usually it unwinds them at a profit. Occasionally the position keeps running against it, and each of the HLP losses described below started that way.
How deposits and withdrawals work
Depositing is a button on the vault page, funded from USDC already sitting in your Hyperliquid account. Withdrawing is the same button in reverse, with a lock-up attached. For HLP the protocol vault docs say you can withdraw four days after your most recent deposit, so each new deposit restarts the four day clock. User vaults carry a one day lock-up.
User vaults also pay the vault leader, the trader who runs the vault, for their work. The leader keeps 10% of each depositor's profit and must keep at least 5% of the vault as their own money at all times, so they lose alongside depositors when trades go wrong. The depositor docs give a clean example. You put 100 USDC into a vault holding 900, so you own 10%. The vault grows to 2,000 with no other flows, you withdraw your 200, and the leader keeps 10 of it as a 10% share of your 100 profit, leaving you 190.
The same docs add that "there may be some slippage as you withdraw and open positions are closed." In plain terms, paying you out can mean the vault has to sell positions at whatever price the market offers in that moment, so you may receive a little less than the balance on screen. If a vault lacks the free margin to cover a withdrawal, it cancels open orders first and then closes 20% of its positions at a time until enough margin is released, per the leader docs. A large withdrawal by someone else can therefore trigger trades that change the positions, and the results, of everyone who stays in.
One structural point sits underneath all of this. No company takes custody of your USDC. Your deposit is pooled with everyone else's in the vault and you hold a share of that pool. You keep the keys to your own wallet, but the vault controls the pooled assets, and getting your money out follows its lock-up and withdrawal rules. That changes the risk profile compared with a lender like Celsius, where customers handed their coins to a company, but it does not remove counterparty risk. You depend instead on Hyperliquid's chain, its validators, the code that runs the vault and the governance decisions behind it, the trade we lay out in our custody guide.
What the APR on the vault page actually measures
The APR shown for any Hyperliquid vault is computed from its recent results. In effect it takes how the vault did over a recent stretch and expresses that as a yearly rate. It looks backward rather than forward, it is not a forecast or a rate anyone has promised you, and for HLP the past is unusually lumpy. An analysis by CoinGecko of the vault's own API data found long flat stretches broken by sharp jumps. The 10 October 2025 crash added an estimated $41.5 million to HLP in a single weekend, roughly a 10% gain for depositors, and the liquidation of one large leveraged ETH long on 31 January 2026 added about $15 million more. Those two events, covering less than two weeks combined, accounted for roughly 41% of the vault's all-time profit.
Over its whole life the result is positive. Hyperliquid's public API, the data behind the vault page, put HLP's cumulative profit at about $138.7 million since its May 2023 launch when we checked in October 2026. Deposits have been far less steady. DefiLlama shows the vault peaking at just over $600 million around the turn of August and September 2025 and sitting near $178 million in mid October 2026.
So read the APR column the right way round. A 3.5% reading says the last stretch was quiet. A much higher reading usually means a crash has just been absorbed. Neither tells you what next month pays, which is the same gap between a headline and an outcome that we covered in our look at USDC rates.
The incidents depositors should know about
HLP's record includes four episodes worth knowing in detail, because each one shows a different side of the vault's job.
On 12 March 2025, a trader opened a 50x leveraged ETH long worth more than $200 million on about $4.3 million of margin, then withdrew collateral until the position fell below maintenance requirements. According to CoinDesk, the trader walked away with a $1.8 million profit while HLP absorbed a $4 million loss. Hyperliquid responded by cutting maximum leverage to 40x on BTC and 25x on ETH.
On 26 March 2025 came JELLY. A trader shorted JELLYJELLY, a thinly traded memecoin, on Hyperliquid, then bought it aggressively on spot markets elsewhere. Hyperliquid values positions using an oracle, a price feed drawn from outside markets, so that buying pushed up the price Hyperliquid saw. The short was liquidated, HLP inherited it, and as the price kept climbing the vault's unrealized loss reached $13.5 million, per CoinDesk's report. Hyperliquid's validators then voted to delist the market. CoinDesk reports that it was settled at $0.0095, against the roughly $0.50 being fed to the oracle from decentralized exchanges, and that the Hyper Foundation promised to make all users whole apart from flagged addresses. Settling at the lower price cut what the inherited short could lose, and CoinGecko's analysis puts HLP's reported final result on the position at a profit of about $703,000. What any one depositor made or lost still depended on when they entered and left the vault. Hyperliquid's delisting rules describe settlement at the one hour time weighted oracle price before the vote, while JELLY was settled at a different price set by validator vote. That precedent cuts both ways.
On 10 October 2025, a day CoinGecko describes as the largest single day deleveraging event in crypto history, roughly $19 billion of positions were liquidated across the market. HLP did what it is built for and collected the estimated $40 million to $41.5 million described above. The same day triggered Hyperliquid's first cross margin auto-deleveraging in over two years, a last resort in which profitable traders' positions are closed against underwater accounts so the exchange carries no bad debt, as the ADL docs describe.
On 12 November 2025 came POPCAT, in what CoinDesk called alleged manipulation. According to blockchain analysts cited by CoinDesk, a trader split $3 million across 19 wallets and used it to build a large leveraged long in the POPCAT memecoin. The trader then placed a roughly $20 million buy order near $0.21, which helped push the price up, and later pulled it. With that support gone the price collapsed, the longs were liquidated and HLP inherited them. POPCAT kept falling, and HLP was left with a $4.9 million loss.
The same day, Hyperliquid paused deposits and withdrawals. The Block reported that a screenshot from the platform described the pause as maintenance, and that an onchain analyst linked it to the POPCAT trade. CoinGecko notes the loss was under 1% of HLP's deposits at the time, and it groups a separate Fartcoin incident in April 2026 with the same pattern.
The losses in these episodes were trading and liquidation losses rather than any reported theft of deposits. They came from the vault doing its disclosed job of absorbing liquidations, sometimes while being targeted for it, and they still reduced the vault's value and with it every depositor's share. The part to weigh is not that HLP sometimes loses, since any liquidity provider does, but that several of its documented losses came from alleged manipulation of thin markets, and that in the biggest one the result was decided by a vote.
The risks the APR does not show
The first is manipulation of small markets. HLP backstops liquidations across a long list of perpetual markets, including memecoins whose spot liquidity elsewhere is thin enough to move with a few million dollars. JELLY, POPCAT and the Fartcoin incident illustrate how exposed the vault is to manipulation in thin markets, and some of that exposure is inherent, because taking over liquidations nobody else wants is the vault's job. If you have read our breakdown of yield farming risks, this is the perpetual futures version of the same lesson, where the yield is payment for absorbing someone else's tail.
The second is governance. Validators can delist a market and choose its settlement, as JELLY showed, which protected depositors once and could just as easily work against them in a different dispute. The third is the plumbing itself, meaning the chain, the bridge and the vault code, which have to keep working for the four day exit to mean anything. And there is no deposit insurance behind any of it. The Foundation's reimbursement after JELLY was a decision taken during the crisis, not a guarantee written into the product.
Who Hyperliquid vaults are best for
HLP genuinely suits someone who already holds USDC on Hyperliquid, is comfortable with a self-custody wallet, and wants exposure to the exchange's own flow without paying a profit share. It suits people who can accept long, quiet months, the occasional large gain on a crash day and the occasional manipulation loss, and who are fine waiting four days to exit. For that person it helps that HLP's positions and vault data are public onchain, though interpreting performance still depends on the data and the method used to read it.
User vaults suit a different reader, someone who wants to back a specific trader and is prepared to judge them on drawdown and history rather than on a trailing APR. The 10% profit share is reasonable for genuine skill, and the leader's mandatory 5% stake gives them a reason to care, but a leader can still lose most of a vault in a bad week, and the APR column will not warn you first.
Neither is a good home for money you may need at short notice, or for anyone who wants a stated rate. If that is you, the honest comparison is with products that publish their terms up front.
How EarnPark compares
If you are comparing HLP with a custodial alternative, the useful questions are what strategy takes the other side of your deposit, what the published rate represents, and what evidence you can inspect. Here is how those answers look for EarnPark.
EarnPark is a custodial yield platform. Deposits go into named strategies, each published in the app with its own risk label, payout cadence and withdrawal terms before anyone deposits.
The EarnPark product that also earns from trading activity is the Alpha Vault family, listed under Advanced Yield with a Market-Exposed risk label on the USDT page. Both products expose depositors to trading activity and market risk, but through different strategies and venues. Per our August report, each Alpha Vault holds roughly half its position in the asset and half in stablecoin, earns from trading fees and liquidations, and rebalances back to that split as prices move. That rebalancing creates a distinct risk in one-way markets. When the price rises, the vault sells some of the rising asset to restore the 50/50 split, so a rising month can end with fewer coins but more dollars, an effect similar in spirit to impermanent loss. In August 2026, a month when prices rose sharply, every Alpha Vault except the USDT one ended with fewer coins than it started with, even though all of them were higher in dollar terms.
As of October 2026, the USDT page reads up to 15% APY, and the top of that range is Alpha Vault, which pays out monthly and processes withdrawals monthly. That is not a direct rate comparison. HLP takes USDC and shows a variable, trailing APR, while the EarnPark figure is the top of a published USDT APY range, coming from a Market-Exposed strategy with monthly payouts and withdrawals. The other strategies on the same page include Maker Core, whose withdrawals go through a bonding period of up to 30 days, and the DeFi and Liquidity Providing strategies, which can be withdrawn anytime. Each published figure is the base any account gets holding zero PARK, EarnPark's token, and EarnPark's rate policy puts no PARK balance check in front of it. HLP is equally open and asks for no HYPE, so the real differences lie elsewhere.
Those differences are custody and verification. With HLP you keep your wallet keys and can inspect every trade onchain. With EarnPark you hold a claim on a custodial platform. EarnPark publishes regular proof of reserves reports prepared by the EarnPark group itself. They give a reserve snapshot, not HLP-style onchain visibility into each trade, and they are not an independent audit. The 10 October 2026 report put 76% of reserves on exchange accounts. Custodial platforms also carry operational risk that depositors absorb, which our risk statement sets out. Read both products' terms before choosing either.
Verdict on Hyperliquid vaults
HLP is a fee free way to take a share of a large exchange's market making and liquidation flow. Its positions and vault data are public onchain, and as of October 2026 it has a positive lifetime record, though that record does not predict any one depositor's outcome. Its weaknesses are just as structural. The APR is a trailing number that swings between quiet and spectacular, several documented losses came from alleged manipulation of thin markets, and the biggest rescue so far depended on a validator vote. User vaults add a skilled or unskilled human to that picture for a 10% cut. Use HLP if you already live on Hyperliquid, hold your own keys and can sit through flat months. If you would rather see a rate, a risk label and the withdrawal terms before you deposit, you can start earning on EarnPark once you have compared both sets of terms yourself.
FAQ
Is the Hyperliquid HLP vault safe? The Hyperliquid HLP vault has no company custodian and its positions are public onchain, but it is not risk free. It absorbed multi-million dollar losses from a whale's leveraged ETH trade and a POPCAT manipulation in 2025, and came through the JELLY squeeze only after validators delisted the market, so treat it as a trading strategy with tail risk rather than a savings account.
Can you lose money in HLP? You can lose money in HLP, because depositors share the vault's losses as well as its profits. Its lifetime result is positive, but individual incidents have cut its value, and anyone who deposits shortly before a loss and withdraws after it takes that hit.
How long is the Hyperliquid vault lock-up? The Hyperliquid vault lock-up is four days for HLP and one day for user vaults, per Hyperliquid's depositor docs. Per the protocol vault docs, the HLP clock runs from your most recent deposit, so adding funds restarts it.
What APR does HLP pay? HLP does not pay a fixed APR. The figure on the vault page is computed from recent results, and it read roughly 3.5% when we checked in October 2026, but crash weekends can add several percent in days while quiet months add very little.
What is the difference between HLP and user vaults? HLP is Hyperliquid's own protocol vault, running market making and backstop liquidations with no fees or profit share. User vaults are run by individual traders who keep 10% of each depositor's profit under Hyperliquid's leader rules, must hold at least 5% of the vault themselves, and can trade only validator-operated perps.
Do you need HYPE to deposit into a Hyperliquid vault? You do not need HYPE to deposit into a Hyperliquid vault. Deposits are made in USDC from your Hyperliquid account, and HLP charges no fee, so the token plays no part in access to the vault.