Verdict
The first decentralised perpetuals venue where execution quality is not an excuse. Fills are fast, the book is real, and every order is publicly visible. The JELLY episode showed the validator set will intervene when the alternative is a loss — reassuring or alarming depending on why you came.
- Best for
- Perp traders who want on-chain settlement without giving up execution
- Cost
- 0.015% maker / 0.045% taker at base tier
What works
- Genuine central limit order book running on its own chain, not an AMM approximation
- Order placement and cancellation cost no gas
- Every position, liquidation and fill is publicly auditable
- HLP vault gives passive users a transparent market-making exposure
What does not
- March 2025 JELLY incident saw validators force-settle a market
- Validator set is small, so 'decentralised' is doing some work in that sentence
- No fiat rails — you arrive with stablecoins or you do not arrive
- Listing of thin markets creates squeeze risk the venue has already been caught by
The pitch for on-chain derivatives has always had a gap in the middle. Self-custody, transparency, no withdrawal risk — and then you place an order and wait six seconds for a block, pay gas to cancel it, and get filled at a price a centralised venue would be embarrassed by.
Hyperliquid built a purpose-made chain to close that gap, and it mostly worked. Orders match in a real central limit order book. Placement and cancellation are free. Latency is low enough that professional market makers actually quote. The result feels like an exchange, which no previous on-chain attempt managed to achieve.
Fees
Base tier is 0.015% maker and 0.045% taker, falling on 14-day rolling volume. That undercuts most centralised perpetual venues on the taker side, and it does so while settling every fill on a public chain where anybody can verify it happened.
There is no gas cost for trading activity. You pay to bridge in and out; between those two points, order flow is free at the network level, which changes what strategies are viable — quoting and cancelling frequently is not punished the way it is on a general-purpose chain.
HLP, and what you are actually underwriting
HLP is the protocol's liquidity vault. Deposit USDC and you take a share of market-making and liquidation profits — and of the losses. It has generally been profitable, which is the correct expectation for a market maker with an edge, and it is emphatically not a yield product.
You are the counterparty to liquidations. In a disorderly market that is a position with real downside, and the vault page is honest about it in a way that most competing 'passive yield' products are not.
The JELLY incident
In March 2025 a trader opened a large short in JELLY, a thinly traded listed market, then pushed the spot price up on other venues. The resulting mark-to-market loss landed on HLP. Hyperliquid's validators responded by delisting the market and force-settling it at a price that limited the damage.
Two readings, both correct:
The system worked. A manipulation attempt was neutralised and passive depositors were protected from a loss they had no way to anticipate.
The system is not what it says on the tin. A small validator set can change the rules of a live market when the outcome is bad enough for the protocol.
If you came for self-custody and transparency, you still have both. If you came because you believed no party could intervene in your position, that belief was tested and it failed. Both facts belong in your model, and the second one is the reason this does not score higher than Deribit.
A venue that can force-settle a market to protect its liquidity pool is safer for depositors and less trustless for traders. You cannot have both, and Hyperliquid chose depositors.
Practical notes
Bridging is the only real friction: bring USDC, and budget time for it on both legs of the round trip.
Thin listed markets carry squeeze risk that the fee schedule does not price. Size accordingly, and check the open interest relative to the underlying's real liquidity.
Everything you do is public. Large positions get watched, copied and deliberately targeted by other users in real time, and there is no hiding a liquidation price.
Verdict
What public positions change
On a centralised venue, your liquidation price is known to you and the exchange. Here it is known to everybody. Large positions are tracked by third-party dashboards, discussed publicly, and occasionally targeted deliberately — a phenomenon that has produced several well-documented episodes of traders being hunted into their own liquidation level.
This is a genuine structural difference rather than a quirk, and it cuts both ways. Transparency means you can see the aggregate positioning of the market before you take a side, which is information no centralised venue will ever give you. It also means you cannot hide, and size that would be unremarkable elsewhere becomes a target here.
Listings and the token
New markets appear quickly, including on assets with thin spot liquidity — which is precisely the condition that produced the JELLY episode. The protocol has tightened listing criteria since, but the underlying tension remains: listing breadth drives volume, and thin markets are where manipulation is cheap.
The HYPE token and its distribution are a separate question from the exchange's quality as a venue, and this review does not score it. Note only that a large share of the venue's early growth was driven by airdrop expectations, which makes historical volume a poor guide to organic demand.
Score: 8.5. The best-executing decentralised derivatives venue available and, on taker fees,cheaper than most centralised ones. It loses points for a validator set small enough to make intervention possible, and for listing markets thin enough to make intervention necessary.