Market Making on Hyperliquid
Market making on Hyperliquid means continuously managing limit orders, inventory, adverse selection, fees, latency, and risk rather than simply placing bids and asks around a midpoint.
A market maker provides executable bids and asks while trying to earn enough spread and rebates to compensate for inventory risk and trading against better-informed flow. Hyperliquid's native CLOB makes the mechanics familiar to exchange market makers, but onchain state, API-wallet design, rate limits, fee tiers, and transparent positions change the operating environment.
The basic quoting loop
A simple maker estimates fair value, chooses bid and ask distances, places limit orders, watches fills, updates inventory, and reprices as the market moves. Real systems add volatility estimates, depth, queue position, correlated markets, funding, latency, and capital constraints.
The strategy earns nothing for displaying quotes that never trade and can lose money quickly if stale quotes trade immediately before a price move.
Inventory is the core risk
Every maker fill creates inventory. If the maker buys from a seller, it becomes longer until it sells elsewhere or hedges. The quoting model usually shifts prices to discourage further accumulation and encourage trades that reduce inventory.
Cross-market hedging can introduce its own basis and execution risk. A hedge that exists theoretically is not useful if it cannot be executed at the required size during stress.
Fees and rebates
Maker economics depend on the current fee schedule, volume tier, staking discounts where applicable, aligned-quote benefits for relevant markets, and any builder or deployer economics.
A strategy with a small gross edge can become negative after fees. Conversely, a rebate can improve economics but should not justify consistently trading at adverse prices.
Infrastructure requirements
Use dedicated API wallets, hard position and notional limits, WebSocket market data, order-state reconciliation, and rate-limit-aware batching. Measure submit-to-acknowledgement and fill-to-hedge latency.
The system also needs a kill switch. If market data is stale, reconciliation fails, or inventory breaches limits, cancelling quotes is usually safer than continuing with an uncertain state.
How to evaluate a maker
Track realized spread relative to a post-fill mark, adverse selection after different time horizons, inventory PnL, fees/rebates, fill ratio, quote uptime, and tail losses. Gross trading PnL alone can hide whether the strategy earns from providing liquidity or merely takes directional risk.