Hyperliquid launched HIP-3 on mainnet in October 2025, introducing a model where any builder can deploy perpetual futures markets without committee approval.
Validators can slash part or all of the stake if a builder feeds manipulated prices, operates a market recklessly, or poses a threat to network solvency. Even during the seven-day unstaking period, the collateral remains vulnerable to slashing.
The protocol burns slashed HYPE rather than distributing it to users, eliminating incentives for false accusations.
Builders control their market’s price oracle and update logic entirely, allowing the listing of virtually any asset.
Still, it introduces oracle manipulation risk, the type of vulnerability that enabled a $112 million exploit on Mango Markets in 2022, where an attacker manipulated a thin price feed to drain the platform.
If a market’s feed fails or a contract expires, builders can invoke a halt function to settle positions at fair value and freeze trading.
The system assumes builders will select reliable oracle sources because their stake depends on it. Validators monitor markets continuously and can slash deployers who use easily manipulated feeds or allow abnormal operation.
Each builder-deployed-market operates as an isolated perpetual exchange with independent order books, margining, and risk parameters. Cross-margining with other assets is prohibited, preventing volatility in one market from contaminating others.
HIP-3 enforces two types of open interest caps. The first consists of notional caps limiting the total dollar value of positions. The second consists of size caps restricting absolute position sizes.
These caps apply per asset and globally across all assets a builder lists. Builders can adjust caps within protocol-defined bounds, but validators expect conservative defaults for volatile or new assets.
Deployers also set leverage limits and initial margin requirements. The framework prevents any new market from becoming systemically critical overnight.
New markets are launched through a Dutch auction that runs every 31 hours. Builders bid HYPE to win deployment slots. To lower entry barriers, the first three markets a builder deploys are auction-exempt.
Beyond winning an auction and staking 500,000 HYPE, no committee approval is required. Any asset can be listed if the deployer backs it with a stake. The protocol includes minimal listing rules.
For example, if a token used as a quote asset for collateral is deemed insecure, validators can vote to revoke its status, automatically disabling markets that use it.
The steep bond requirement implicitly filters for serious projects with sufficient capital and expertise. Hyperliquid’s documentation states the goal is to ensure “high quality markets and protect users” from temporary listings.
dYdX v4 is transitioning toward permissionless markets but still requires governance votes for new listings. The platform plans to implement an isolated margin for risky assets and enforce strict oracle requirements. Assets must trade on at least six major exchanges to ensure robust price feeds.
Chaos Labs proposed a “probationary asset” phase with separate insurance funds and tighter trading bands for new markets.
GMX v2 addresses similar concerns through isolated liquidity pools per trading pair and Chainlink oracles for pricing. The platform integrates Chaos Labs’ Edge Risk Oracle system, which dynamically adjusts open interest caps and price impact coefficients based on real-time conditions.
Additionally, each GMX market is ring-fenced, as issues in one pool don’t affect others.
Drift Protocol on Solana utilizes Switchboard’s permissionless oracles to list new assets rapidly, but enforces a 10% circuit breaker band.
If the mark price diverges from the oracle’s five-minute time-weighted average by more than 10%, the market prevents new orders outside that band. Drift also limits single trades to 2% price impact maximum.
During the HIP-3 evaluation phase on testnet, no significant issues were reported. A $21 million theft from Hyperliquid around the same timeframe was a private key compromise unrelated to market operations, resulting from user operational flaws.
The protocol’s true test will come when third-party builders deploy novel markets for exotic indices or real-world assets.
HIP-3’s design combines economic deterrence through staking with technical constraints through caps and isolation. Validators serve as a final backstop, able to slash up to 100% of the stake for violations threatening network correctness or solvency.
The architecture effectively transforms Hyperliquid into financial infrastructure rather than a single exchange. Each new market functions as its own mini-exchange secured by the network.
The answer is layered. Builders keep markets safe because their capital is at stake. Validators maintain market safety through monitoring and slashing authority. The protocol maintains market safety through automated caps, isolation, and sanity checks.
This model assumes rational actors that a $20 million bond will deter manipulation more effectively than committee gatekeeping. It assumes that validators will act when needed, but the system itself is robust enough that slashing should “never” be necessary on the mainnet, as Hyperledger’s team stated.
Lessons from Mango and GMX directly informed these safeguards. Whether the combination of stake, isolation, and oversight can handle all edge cases remains to be proven through live markets.
For now, Hyperliquid offers a straightforward proposition: any asset can become a perpetual market if someone believes in it enough to risk $20 million.
The protocol bets that price is high enough to separate serious builders from reckless ones, and that layered defenses can catch what economic incentives miss.
