Hyperliquid HLP: What the House Actually Pays (and the Days It Doesn't)

Every trader who has been stopped out on a centralized exchange has heard the same explanation. The market moved. The book was thin. The house always wins. You take the loss, the market-maker takes the spread, and the liquidation engine takes whatever was left of your margin. The house's edge isn't a secret — it's just never been for sale.

Hyperliquid is the first perp exchange at scale to put that edge up for sale. Instead of routing order flow to a proprietary market-making desk, it splits the house's job in two and hands both halves to the community. Market-making and liquidation backstopping are consolidated into a single on-chain vault called HLP — the Hyperliquidity Provider vault. Deposit USDC and you are underwriting the same book that fills your orders, sharing in the same spreads you would otherwise pay.

The headline economics are genuinely unusual for DeFi: no performance fees, no management fees, and a track record of roughly $122 million in cumulative profit since inception — about 450% total return as of early 2026, on a venue now clearing more than $10 billion in daily volume.

But "the house always wins" is a slogan, not a schedule. HLP has months where it makes +6.4% and months where it makes 0.07%. It has had a single day it made ~5.8%, and it has had a week in 2025 when the whole experiment nearly went sideways. This article walks through where the yield actually comes from, what the numbers show, and what you are really signing up for.

Where HLP's yield actually comes from

HLP is a pool of USDC owned by depositors and managed by Hyperliquid's automated vault strategy. It does three jobs at once, and each job is a separate revenue stream:

The fee context matters here. Hyperliquid charges traders 0.045% per taker fill and 0.015% per maker fill — among the lowest rates in perps, as we detail in our dYdX vs GMX vs Hyperliquid comparison. HLP sits on the maker side of that book: when you cross the spread for instant execution, the spread itself and a share of the fee flow land in the vault. Volume is the tide that lifts this boat, which is why 2026's $10B+ daily volume regime has been so kind to it.

Notice the pattern, though: two of the three streams are paid by exactly the traders described in the opening of this article. If you trade perps on Hyperliquid and hold HLP, you are on both sides of your own trade. (New to leverage trading entirely? Our decentralized perps guide covers the mechanics first.)

What HLP has actually paid — the honest numbers

Marketing pages show you the since-inception chart. You should see the dispersion too. Both, side by side:

The February 1 event deserves a closer look, because it is the vault's business model in miniature. A leveraged long position worth roughly $700 million was liquidated in a cascade. The open orderbook couldn't absorb the flow, so HLP's backstop stepped in and took the position. Markets recovered, the vault resolved the position, and it booked about $15 million in a single day — roughly +5.8% in 24 hours. If you want to understand why those cascades create this opportunity, we break down the mechanics in how liquidation cascades work.

Now hold that against February as a whole: +0.07%. The vault that made 5.8% in a day made a rounding error over thirty days, because nothing got liquidated violently enough to feed the backstop and spreads stayed tight. That is the shape of this yield: good average, violent path. The low-20s% trailing CAGR is real, but it is assembled from a handful of chaotic days and long stretches of near-nothing. Anyone modeling HLP as smooth monthly income is modeling the wrong asset.

HLP vs JLP vs GLV: the house, three ways

Hyperliquid isn't the only protocol selling a seat at the house's table. Jupiter's JLP pool does the same job on Solana, and GMX's GLV pools do it on Arbitrum. If "be the house" appeals to you, the real question is which casino:

Three differences are worth your attention. First, fees: HLP takes nothing from depositors, while most competitors either charge performance fees or route a share of platform revenue elsewhere — over years, zero-fee compounding is a quiet advantage. Second, composition: HLP's yield leans harder on the liquidation backstop than JLP or GLV, which makes it more sensitive to volatility regimes — higher highs, quieter lows. Third, underlying volume: all three are derivatives of trading activity, and Hyperliquid's $10B+/day is currently the strongest of the three venues.

Whichever pool you pick, the intellectual framework is the same one behind delta-neutral yield farming: you are selling insurance to leveraged traders and collecting premium. It works until the claims all arrive at once.

The risks — including the day the house almost lost

In March 2025, an attacker demonstrated the vault's structural weakness with surgical precision. By holding both a long and a short in JELLY on separate venues, then pumping the price, the attacker forced HLP to absorb a ballooning delisting-bound position through the backstop. The vault briefly carried an $11 million unrealized loss, with worse feared if the squeeze continued. What happened next is the part you must understand: Hyperliquid's validators voted to delist JELLY and settle the position at fair value rather than let the market run. HLP ended up fine — but a social consensus intervened in a market outcome, and that is a genuine centralization data point, not a hypothetical. If you deposit, you are trusting that discretion to be exercised benignly next time.

To the team's credit, the response was structural: open-interest caps on low-liquidity assets, improved auto-deleveraging (ADL) parameters, and position-concentration monitoring designed to make a repeat of the JELLY playbook much harder. The risk has been reduced. It has not been eliminated — no governance patch ever eliminates tail risk; it just moves where you have to look for it.

• Adverse selection. The backstop only gets the trades nobody else wanted. You profit when markets recover and bleed when they don't — the classic insurance trade, executed in public.
• Drawdown months. +6.40% and +0.07% are both real recent months. Negative months exist too. If you would redeem at the first red month, the asset doesn't fit you.
• The 4-day lock is not stop-loss-able. During a lockup you cannot exit, whatever the chart does. Size as if the exit button doesn't exist for four days — because it doesn't.
• Platform concentration. HLP lives entirely on Hyperliquid's own L1 with its limited validator set and the newer HyperEVM stack. A chain-level incident is a vault-level incident. (This is part of why we grade Hyperliquid B, not A, in our perp platform comparison.)
• Not a savings account. Nothing here is a deposit or a bond. This is an active trading strategy you own a slice of, and none of it is financial advice.

For context, here is how the lockup and fee structure compares across the ways to put capital to work on Hyperliquid:

So should you deposit?

HLP is one of the most honest products in DeFi, paradoxically because its economics are so transparent: you are paid to provide the liquidity and absorb the liquidations that other traders create, and you keep all of it. Over full cycles, that trade has paid roughly a low-20s% CAGR lately and ~450% since inception. In exchange, you accept violent return dispersion, a 4-day lockup you cannot breach, and a protocol whose governance has already shown — once, under fire — that it will override markets to protect the vault.

If that trade sounds fair to you, size it like an insurance book, not a bank deposit: money you won't need for weeks, in an amount whose worst month wouldn't change your behavior. If it doesn't, that's a perfectly good answer too. Either way, run the numbers before you commit — our yield-risk grader scores HLP against your own return requirements and risk tolerance.

Frequently asked questions

What is the Hyperliquid HLP vault?

HLP (Hyperliquidity Provider) is Hyperliquid's flagship community vault. Depositors pool USDC that the vault deploys as market-making liquidity on the orderbook, as liquidation backstop during cascades, and as lending supply earning USDC interest. There are no performance fees, no management fees and no deposit or withdrawal fees. It has accumulated roughly $122 million in profit since inception — about 450% total return as of early 2026.

How much yield does HLP pay?

Returns are strong but volatile. The trailing 12-month CAGR was in the low-20s% as of Q1 2026, within a typical 10–25% band. Monthly results swing widely: January 2026 returned +6.40% while February 2026 returned +0.07%. On February 1, 2026, a single $700 million liquidation cascade handed the vault about $15 million in one day — roughly +5.8% in 24 hours.

What are the risks of depositing into HLP?

The main risks: adverse selection (the backstop absorbs toxic positions precisely when markets are most violent), highly variable monthly returns with drawdown months, a 4-day lockup during which you cannot exit, platform risk on Hyperliquid's own L1 and limited validator set, and governance discretion — demonstrated in March 2025 when validators voted to delist JELLY and settle it at fair value while HLP briefly carried an $11 million unrealized loss. HLP is not a savings account and this is not financial advice.

How does the 4-day withdrawal lock work?

Every deposit starts a 4-day lockup counted from that most recent deposit. During those 4 days you cannot withdraw. Deposits and withdrawals themselves are free, with no performance fees — and adding funds resets the clock on the newly added amounts. For comparison, Hyperliquid user vaults lock funds for 1 day but charge leaders a 10% performance fee, and HYPE staking requires a 7-day unbonding period.

Sources and further reading

• Hyperliquid docs — official documentation on HLP, vault mechanics and lockup rules.
• DefiLlama — Hyperliquid HLP — live TVL, fees and yield tracking for the vault.

Related reading: The Decentralized Perps Guide, How Liquidation Cascades Work, Delta-Neutral Yield Farming and our dYdX vs GMX vs Hyperliquid comparison.