How to Set Up a Trading Bot on Hyperliquid

Hyperliquid is an on-chain perpetuals exchange with an order book, which makes it unusual: you get centralised-exchange execution semantics with self-custody. That combination changes how bot setup works compared with Binance or Bybit, and the differences are exactly where first-time setups go wrong.

• Hyperliquid authenticates with an EIP-712 signature from an agent wallet, not an API key and secret.
• The agent wallet can trade but can never withdraw — that is enforced by the protocol, not by a settings toggle.
• If you trade from a sub-account or vault, your software must be told the master address or every order fails.
• Reduce-only limit orders below roughly $10 notional are rejected, which silently breaks multi-leg exit ladders on small positions.
• There is no gas cost per order, so ladder strategies that would be uneconomic on other chains are viable here.

Why run a bot on Hyperliquid?

Three properties matter for automation specifically:

• A real order book. Not an AMM. Limit orders, post-only, reduce-only and stop triggers behave the way a bot expects, so strategies written for centralised venues port over.
• Self-custody. Your funds sit in a protocol account you control. There is no exchange withdrawal queue and no counterparty deciding when you can leave.
• No per-order gas. Orders are sequenced by the chain without a per-transaction fee to you, so a ten-leg entry ladder costs nothing extra to place. On a general-purpose L1 the same strategy would be dominated by gas.

The trade-off is a smaller pair universe than Binance and thinner books on the long tail. For majors it is competitive; for a low-liquidity altcoin, check the book before you automate anything against it.

Step 1 — Fund your Hyperliquid account

Deposit USDC via the bridge from Arbitrum, or use one of the supported deposit routes in the interface. Wait for the balance to appear in your perps account rather than spot — bots trade the perps account, and a balance sitting in spot will look like an empty account to your software.

Start with an amount you would be comfortable losing entirely while you verify the setup behaves as expected.

Step 2 — Create an API wallet (agent)

This is the step that differs most from other exchanges. Hyperliquid does not issue an API key and secret. Instead you authorise a separate wallet — an agent — to sign actions on behalf of your main account.

1. Open the Hyperliquid interface and go to the API section of your account settings.
2. Generate a new API wallet. You will be shown a private key.
3. Authorise it. This is an on-chain action signed by your main wallet, and it is what links agent to account.
4. Copy the private key somewhere safe. It is shown once.

The security property that matters: an agent wallet can place and cancel orders. It cannot initiate a withdrawal. That is a protocol-level constraint, not a permission checkbox that could be misconfigured. A leaked agent key lets an attacker trade your account badly — which is bad — but not drain it.

Step 3 — Connect the agent to your software

Paste the agent private key into your platform's credential form. In TradeFloor that is Settings → Credentials → Add credential → Hyperliquid; the key is encrypted at rest and order signing happens in an isolated component that never exposes it back to the application.

The sub-account and vault trap

If your agent was created under a sub-account, or you trade through a vault, orders must be submitted with the master account address attached. Omit it and every order is rejected with an authorisation error that reads as if the key itself is wrong — which sends people back to regenerate a key that was fine all along.

Good platforms detect this automatically by querying the wallet's role and pinning the master address for you. If yours doesn't, find the field called vault address, master address or account address and fill it in. This is the single most common first-time setup failure on Hyperliquid.

Step 4 — Configure your first strategy

Resist the urge to start with something clever. A first configuration should be boring enough that you can predict exactly what it will do.

The $10 notional floor

Hyperliquid rejects reduce-only limit orders below roughly $10 of notional value. This has a non-obvious consequence for exit ladders. Split a $40 position across five take-profit legs and each leg is $8 — under the floor. The first legs may place while the last ones fail, and unless your software validates the whole ladder up front, you get a position whose actual exit plan does not match its configuration.

Two rules follow: keep early positions to a single take-profit leg, and when you do ladder, check that position_notional / number_of_legs clears $10 with margin to spare.

Insurance and averaging orders

The same floor applies to averaging ladders, where it compounds — each rung is multiplied by your martingale factor, so the smallest rung is the one at risk. Before enabling averaging, read how insurance orders work and where martingale sizing breaks.

Step 5 — Verify the position is actually protected

After the first entry fills, do not assume the bracket exists. Check it, in this order:

1. Open the Hyperliquid orders panel directly, not just your platform's view.
2. Confirm a reduce-only stop order exists on the correct side, at the price you expect.
3. Confirm the take-profit is there and that its size matches the position.
4. Confirm the total reduce-only size does not exceed the position size.

A live position without a live stop is the most expensive state a bot can be in, and it is invisible unless you look. Any platform worth using also runs an automatic reconciliation pass that re-places missing protective orders continuously — but verify it manually the first few times so you know what correct looks like.

Step 6 — Monitor, then scale slowly

Watch the first ten to twenty trades before changing anything. What you are looking for is not profit; it is correctness:

• Did entries fire where your rule said they should, and not fire where it said they shouldn't?
• Did stops execute at roughly the configured distance, or was slippage materially worse?
• Did anything close for a reason you did not configure?
• Does the realised cost per round trip match what you assumed when you designed the strategy?

Only once all four are boring should size go up, and then in steps small enough that a bad step is recoverable.

Common Hyperliquid setup errors

How Hyperliquid compares for automation

Against Binance and Bybit, Hyperliquid trades pair breadth for custody and predictable order semantics. The authentication model is genuinely different — signature-based rather than key-and-secret — which matters if you are writing your own client. That comparison is broken down properly in the API comparison for algo traders, and the trader-level view is in Binance vs Bybit vs Hyperliquid.

// questions or corrections · [email protected] · more essays · /blog