Separate the risk layers

A wallet-based exchange changes who controls assets, but it does not remove protocol failure. Hyperliquid’s own risk documentation lists bridge smart-contract, L1, liquidity and oracle risks and notes that the list is not exhaustive.

Add user-controlled risks: compromised signing devices, wrong-chain deposits, malicious front ends, oversized leverage, crowded funding and stop slippage. Each layer needs a different control.

LayerExamplePractical control
KeySeed or signer compromiseDedicated wallet; hardware signing
BridgeContract or transfer failureTest transfer; limit parked capital
ChainValidator or liveness eventSize exposure; monitor status
MarketThin depth or oracle stressLimit size; stress slippage
PositionLeverage and liquidationPredefined loss; margin buffer

Self-custody is a responsibility transfer

You control signing, but funds deposited into the trading system still depend on protocol infrastructure. A hardware wallet protects a key from some attacks; it does not validate an economic decision or guarantee the front end is genuine.

Use bookmarks, verify domains, read transaction intent and separate trading capital. No legitimate support agent needs a seed phrase.

A conservative operating policy

This is not a claim that the platform is safe or unsafe. It is a framework for deciding which failure modes you can tolerate and how much capital belongs in each layer.

  • Deposit only capital assigned to active trading.
  • Test deposits and withdrawals.
  • Use a separate wallet and revoke unrelated approvals.
  • Keep leverage below the point where normal volatility threatens liquidation.
  • Treat displayed APRs, funding and past vault returns as variable.
  • Have an exit path that does not depend on one interface.