HLP vs JLP vs GLP — same $10,000, illustrative bands
Representative APR/drawdown ranges from public vault history, not live-fetched. Confirm current numbers on each protocol's dashboard before depositing.
| Vault | APR range | Max drawdown | Yield/drawdown ratio | Lockup |
|---|
You're the house — what that actually means
A vault APR is not a fixed coupon. It's the net result of spread capture, funding received, and trader P&L flowing through the book — mostly positive on quiet, choppy days, and occasionally sharply negative when a large trader is right on a directional move. That's structurally different from lending or staking yield, where the downside is a slow bleed, not a fast one. Model the correlated-stress case: if the historical worst drawdown repeats while you're deposited, does the excess yield over the last few months still leave you ahead? See stablecoin vault risk for the equivalent framing on stablecoin yield vaults.
Why HLP, JLP and GLP need their own risk math
Every staking or lending calculator on the internet treats yield as roughly linear and the downside as slow — miss a payment, get slashed a known percentage, done. Perp-DEX liquidity vaults don't work that way. HLP is Hyperliquid's protocol-owned market-making vault: deposit USDC, become a fractional counterparty to every perp trade on the exchange, and earn the spread, funding and losing-trader P&L that flows through the book. Historically that's averaged somewhere in the 15-30% APR range across most quarters — until a single crowded trade goes hard against the book. The March 2025 JELLY incident is the textbook case: a trader's outsized short on a thin token pushed HLP's mark-to-market loss on that one position past 10% of vault assets in hours, before the exchange intervened. That's not a slow bleed. It's a fast one.
JLP on Jupiter (Solana) is structurally different again — it's a fixed-weight basket of SOL, ETH, WBTC, USDC and USDT that earns trading fees and net trader losses, but is not delta-neutral. Holders carry the basket's own directional exposure on top of the counterparty risk, so a broad market downturn hits JLP twice: once as falling collateral value, once as traders statistically doing better in a trend. GMX's GLP running on Arbitrum and Avalanche sits somewhere in between — a similar basket model with historically lower but steadier APR and a much shorter effective cooldown than HLP's 4-day lock.
The number that actually matters for sizing a deposit isn't the headline APR — it's the yield-to-drawdown ratio: how many "worst historical drawdown" events would it take to wipe out a year of yield. A vault yielding 22% APR with a 10% worst drawdown has a 2.2x ratio; the same 22% with a 20% drawdown has just 1.1x — half as much cushion for the exact same headline return. Pair this with the lockup: HLP's 4-day withdrawal delay means you can't react instantly if a drawdown event starts, a constraint a simple APY number never shows.