Section IV: The HLP Design

With tradable products in place, Hyperliquid faced another challenge: ensuring the liquidity depth necessary for these markets to function. Technical performance alone doesn't guarantee success; traders demand deep liquidity, tight spreads, minimal slippage, and reliable liquidation mechanisms, requirements that have historically favored centralized exchanges with dedicated market makers. Hyperliquid's solution creates new compromises between liquidity provision and risk concentration.

The Hyperliquidity Provider (HLP) represents Hyperliquid's most innovative design choice: a community-owned vault that simultaneously provides market-making services and handles liquidations. Depositors contribute capital to HLP and share in its profit and loss, creating a decentralized market-making system that doesn't rely on external firms. HLP's profits come primarily from market-making spreads and liquidation fees, while losses stem from being on the wrong side of trades against sophisticated traders who possess better information or faster execution, as well as from holding losing positions as the counterparty to winning trades.

HLP's design solves several problems at once. It provides consistent liquidity across all markets, handles liquidations efficiently (crucial for leveraged trading), and distributes market-making profits to the community rather than extracting them to external firms. The system internalizes much of the trading flow, reducing the need for external counterparties.

However, this concentration creates meaningful risks. During extreme volatility, HLP depositors bear the losses from trading against better-informed counterparties and from liquidation cascades. It is also worth noting that HLP now accounts for a small minority of daily trading volume, roughly 1-2%, as organic market-maker participation has grown. While HLP isn't the sole counterparty on the central limit order book (anyone can post liquidity), it provides core baseline liquidity across markets and performs liquidations, creating concentration risk that traditional market-making structures distribute across multiple firms.

The JELLY Manipulation

The JELLY manipulation in March 2025 demonstrated how vault-based systems can suffer losses from coordinated attacks. Attackers opened large leveraged positions ($4.5M short, two $2.5M longs) on a low-liquidity token JELLY, then manipulated the liquidation process while simultaneously pumping the token's price 250% on Solana. This created a $12 million unrealized loss that threatened the protocol's solvency. Validators had to make an emergency intervention, overriding the oracle price to prevent collapse, while the team quickly implemented fixes including better position size limits, improved liquidation mechanisms, and enhanced governance controls. All traders were compensated, but the incident exposed significant vulnerabilities in the platform's risk management architecture.

The October Liquidation Cascade

The October 10 liquidation cascade was the largest stress test of infrastructure. Over $20 billion in leveraged positions were liquidated, and total outstanding perpetual contracts across the market collapsed by 43%. Hyperliquid processed roughly half of all liquidations. Its outstanding positions fell from $14 billion to $6 billion while maintaining 100% uptime and avoiding bad debt.

HLP performed as designed. When the order book couldn't absorb forced selling, HLP's liquidation child vaults stepped in as backstop liquidators. They took over distressed positions plus collateral and unwound into the rebound. Across the event, HLP generated over $40 million in profit, roughly a 10% daily return on vault capital. This was driven largely by liquidation flow during the cascade.

The violence also triggered Hyperliquid's cross-margin auto-deleveraging mechanism (the backstop process explained in Chapter VI) for the first time in over two years. HLP is compartmentalized into child vaults with isolated risk limits, so some liquidation vaults themselves approached negative equity as they warehoused junk exposure. At that point, after regular liquidations and HLP takeovers were exhausted for those specific accounts, the risk engine began trimming profitable positions on the winning side. It used a documented ranking system to keep every margin account non-negative by forcibly closing the most profitable positions first. The result: LP capital backstopped the system and still booked outsized net returns with no bad debt. This occurred even though some profitable shorts, including large external traders and HLP child vaults, were forcibly reduced at locally optimal prices.

Different vault designs absorbed tail risk differently. On Lighter, a competing perpetual DEX examined in Section VII, the LLP vault provides most day-to-day liquidity. When the market gapped and its infrastructure suffered multi-hour outages, LLP took the hit directly. It suffered roughly a 5.3% drawdown, about $21.5 million in losses, while traders unable to manage positions lost $25 to $30 million more. Lighter later compensated via a points program. The structural contrast is clear: HLP is a high-capacity, compartmentalized backstop that turns volatility into yield, while LLP functions as always-on inventory taking first-loss risk.