How Hyperliquidity Provider (HLP) Works:
Mechanics, Risks & Real Numbers
A $400M protocol vault, a 4-day lockup, and what JELLY plus FARTCOIN revealed about the real tail risk.
HLP is the single largest vault on Hyperliquid and the one most people ask about first. It is also the one with the least-understood mechanics and the most-misread risk profile.
This post breaks HLP down without hand-waving: what it actually does, where the yield comes from, what the 4-day lockup means in practice, and why one memecoin in March 2025 cost the vault around $12M.
TL;DR
- Deposit USDC → get a proportional share of a $400M vault.
- Vault earns from three streams: multiple market making strategies, liquidations, and supplying USDC in Earn. It also accrues a portion of Hyperliquid's trading fees.
- Deposits carry a 4-day lockup measured from your most recent deposit.
- HLP is fully community-owned — depositors share the vault's PnL, there is no separate operator cut.
- It is not risk-free. JELLY (March 2025, ~$12M) and FARTCOIN (April 2026, ~$1.5M) both dumped toxic positions on HLP via different attack vectors.
What HLP actually is
HLP stands for Hyperliquidity Provider. Unlike a leader vault run by an individual, HLP is a protocol vault operated by Hyperliquid itself.
The official docs put it plainly:
Its job is to be the counterparty of last resort. When a trader places an order and no natural counterparty is there at that price, HLP can fill the other side. In exchange, it keeps what market makers keep: spread, funding, and the premium collected when positions get liquidated.
Mechanically it is non-custodial and community-owned. USDC sits in a smart contract on Hyperliquid's L1. Depositors share the PnL directly — there is no management fee and no operator skimming the top.
Where the yield actually comes from
Three revenue streams, not one:
| Stream | What generates it | Character |
|---|---|---|
| Market making | Bid/ask spreads and funding payments from providing liquidity on perps | Steady, grinds in most regimes |
| Liquidations | Liquidation premiums when HLP takes over liquidated positions | Spiky — pays big on volatile days |
| Earn on USDC | Interest on idle USDC sitting inside the vault | Boring — but always-on |
Per the official docs, HLP "accrues a portion of trading fees" — so maker, taker, funding and liquidation premiums partly feed the vault, alongside the direct PnL from market making and liquidations. Depositors share that pool proportionally.
This is why HLP's yield tracks trading activity. Quiet weeks = thin returns. Volatile weeks with lots of liquidations = fat weeks. It is closer to a volatility-selling business than a fixed-yield product.
Real HLP numbers right now
Here is what the vault looks like at the time of writing, pulled from our live index:
| Metric | Value |
|---|---|
| TVL | ~$399.8M |
| Age | ~1,078 days (about 3 years) |
| 30-day return | −7.3% |
| 90-day return | +32.8% |
| Risk Score (VaultVision) | 64 · moderate band |
Note the shape: a current 30-day drawdown sitting on top of a very healthy 90-day return. That is classic HLP. The vault does not yield in a straight line — it monetizes flow, and flow is lumpy.
See the live version on the HLP vault page with full risk breakdown, drawdown chart, and position context.
The 4-day lockup, in practice
Every deposit into HLP carries a 4-day lockup measured from your most recent deposit. From the official docs: a deposit made on September 14 at 08:00 becomes withdrawable on September 18 at 08:00. Top up in the meantime and that clock resets for the whole position.
It is a design choice, not an oversight:
- Market makers cannot run if inventory can leave instantly. Forced-to-sell-at-tick is how you blow up a book.
- A locked float lets HLP hold positions through noise without cascading withdrawals.
- It is also why the vault can genuinely act as counterparty on large trades — there is committed capital behind the quotes.
Practical consequence: HLP is not a parking spot. Do not deposit what you might need this week. Size it like an illiquid allocation, not a savings account.
The tail-risk file: JELLY and FARTCOIN
HLP's real risk is not a bad week of market making. It is a toxic position dumped onto the vault by the liquidation engine. That has happened twice, in very different ways.
JELLYJELLY — March 2025
An attacker opened a $4.1M short on JELLYJELLY (a micro-cap memecoin listed on Hyperliquid perps) and simultaneously pumped the spot price across other venues. Within roughly an hour JELLY was up more than 400%, the short was deeply underwater, and the position began cascading through Hyperliquid's liquidation pipeline.
Because of how Hyperliquid inherits distressed positions, HLP ended up holding the short. With ~$230M in the vault at the time, a further squeeze could have been catastrophic.
Hyperliquid's validators reached consensus in roughly two minutes and force-delisted JELLYJELLY, settling the positions at the attacker's entry rather than the pumped market price. HLP's realized loss was limited to about $12M. The attacker walked with a partial profit before withdrawals were frozen.
FARTCOIN — April 2026
Same category of attack, different surface. An entity spread across four freshly-funded wallets accumulated a $15M long in FARTCOIN (145M tokens), pushing price roughly 27% higher ($0.16 → $0.25) in the process. Then the wallets triggered what analysts described as a "suicide liquidation" in thin liquidity.
This time the vector was Auto-Deleveraging (ADL) rather than a validator-level delist. The toxic position was pushed onto HLP through the ADL mechanism, and the vault ate ~$1.5M in absorbed losses. Short traders on the other side were randomly rewarded by ADL with roughly $849K in profits. The wallets booked a $3M paper loss but likely netted out via cross-venue hedging.
What both incidents tell you
- HLP has real tail risk. Any newly-listed perp is a potential attack vector. That risk does not show up in 30-day APR.
- The loss surfaces change, but the pattern repeats. JELLY exploited validator-level position inheritance. FARTCOIN exploited ADL. Future vectors will find new edges.
- Resolution is not always automated. JELLY was a manual validator intervention, which was the right call at the moment but highlights that the decentralization story has caveats.
After JELLY, Hyperliquid tightened leverage caps (BTC to 40x, ETH to 25x) and overhauled listing rules so HLP has more buffer. FARTCOIN showed those changes reduced blast radius but did not close the surface entirely — $1.5M vs $12M is progress, not a solved problem.
Who HLP actually suits
Think of HLP the way you would think about running a market-making desk. You want:
- A time horizon longer than the 4-day lockup — ideally months.
- Tolerance for 5-15% drawdown stretches in exchange for the longer-run yield.
- Position sizing that accounts for tail events like JELLY and FARTCOIN, not just the 1-year chart.
It is a reasonable core holding for a Hyperliquid-native portfolio. It is not a T-bill replacement.
HLP vs individual leader vaults
A question we get often: why not just pick a hot leader vault with 100%+ APR?
| HLP | Leader vault | |
|---|---|---|
| Operator | Protocol | Individual |
| Capacity | Very high ($400M+) | Usually capped |
| Strategy | Market make & liquidate | Directional / stat-arb / copy |
| Drawdown profile | Grinding, with occasional shocks | Can be anything — including blow-ups |
| Key risk | Systemic listing / validator | Leader judgement & skill decay |
They are different tools. HLP is a beta product — you are levered long "Hyperliquid as a venue". A good leader vault is a specific alpha bet. Most portfolios use both.
See HLP live, with full risk breakdown
TVL, rolling drawdown, risk components, and where HLP sits relative to the rest of the Hyperliquid vault universe.
Open HLP vault pageRelated reading
- How VaultVision's Risk Score is calculated
- How to choose a Hyperliquid vault: 5-factor framework
- Top Hyperliquid vaults ranked by risk-adjusted return