How market making works on Hyperliquid: spreads, maker rebates and inventory risk

How does market making work on Hyperliquid? A market maker rests buy and sell orders on both sides of the order book and earns the gap between them, the spread, when both sides fill. On Hyperliquid those resting orders pay a maker fee of 0.015% at the base tier, and accounts that make more than 0.5% of 14-day maker volume earn a rebate instead. The risk is being filled on the wrong side of a move. On Nov 11, 2026, BTC’s spread was 0.14 bps, less than the 1.5 bps base maker fee, so quoting the top of the book on the biggest markets is a fee-tier game.

Spreads and depth on Hyperliquid right now

Seven markets: the four largest, then markets ranked about 10th, 30th and 80th by open interest. Depth is the dollar value resting in the 20 price levels the API returns on each side, and the span is how far those levels reach from the mid price.

The majors are deep and tight: ETH had $8.2M of bids within 9.5 bps of the mid. Further down the list, spreads widen and depth thins quickly, which is where a quoting strategy can earn more per fill and carries more risk per fill.

Trade at maker rates without running a quoting bot

Quoting both sides of the book takes a server, fast cancels and enough volume for a rebate tier. If what you want is the maker side of the fee schedule on your own strategies, Senpi gets you there without a bot to run. Senpi publishes this guide; here is what it does:

• Maker-first entries. Senpi’s runtime places strategy entries as maker orders first, at 0.015% at the base tier instead of 0.045% for a taker fill, and switches to a taker order if the maker order hasn’t filled in time. See fee-optimized execution.
• 0% Senpi fee on paid plans (0.03% on the free trial), so Hyperliquid’s own maker and taker fees are the trading fees you pay.
• Rules instead of quotes. Strategies take positions on a rule, such as a trend, a reversion or a funding extreme, so there is no two-sided inventory to manage.
• A stop from the first fill. Strategies built from a rule or a template carry the Dynamic Stop-Loss (DSL), and each can have its own daily-loss or drawdown halt.
• No server, no key. Each strategy trades from its own strategy wallet on Senpi’s runtime; the agent holds no key and cannot withdraw.

Set it up in the Senpi app, or from Claude or ChatGPT through Senpi Connectors. A strategy you confirm runs as code, and by default no AI model makes live trading decisions. Copy strategies carry at most a flat stop. If you do want to quote both sides of the book yourself, the tools for that are further down.

How a market-making bot works

1. 01Estimate a fair price. Usually the mid, a weighted mid, or a price from a deeper venue.
2. 02Quote both sides around it. Wider quotes fill less often and earn more per fill.
3. 03Size and skew by inventory. As the bot gets long, it lowers its bid and ask to sell more and buy less; models such as Avellaneda–Stoikov formalise this.
4. 04Cancel and replace as the market moves. Stale quotes are what informed traders hit.
5. 05Cap risk. Maximum inventory, a hedge on another market, or a stop that flattens the position.

The fee maths

If a buy and a sell both fill as maker orders at the best bid and ask, the round trip earns the spread and pays two maker fees. At the base tier that is 2 × 1.5 bps = 3 bps.

That is why the tightest books are dominated by large makers with rebates, often including HLP, Hyperliquid’s own community vault, which also quotes markets (Hyperliquid vaults and HLP). A smaller maker either needs volume high enough for a better tier or quotes wider, less crowded markets and accepts more inventory risk. Fees in detail: Hyperliquid fees.

What can go wrong

• Adverse selection. Your quotes fill when the price is about to move through them, so the filled side is often the losing side.
• Inventory. A run of fills on one side leaves a directional position. With leverage, that position can be liquidated.
• Rate limits. Hyperliquid allows 1 action per 1 USDC traded per address after a 10,000 starting buffer. Cancels get a higher limit, so you can still pull quotes, but a bot that churns orders without fills will be throttled.
• Open-order caps. The default is 1,000 open orders per user; above that, new reduce-only and trigger orders are rejected.
• Congestion. During high traffic, an address may use at most twice its previous day’s maker share of block space.

Tools for market making on Hyperliquid

Hummingbot has Hyperliquid spot and perp connectors, market-making strategies and a paper-trading mode. NautilusTrader and CCXT suit a market maker you write yourself.

This is educational material, not financial advice. The market readings are data, not trade signals, and leveraged perps can lose money quickly.