Custodial vs non-custodial copy trading

Four architectures, four failure modes

Almost every copy-trading offer falls into one of four structures, and each fails differently when the operator disappears.

• Pooled custody — you deposit into a shared treasury; recovery depends entirely on the operator.
• Segregated custody — funds are held for you but still by them; withdrawal is a request.
• Unscoped API access — funds stay with an exchange but the key may permit withdrawal or transfer.
• Scoped on-chain agent — trading permission only, revocable by you, funds never move.

What each model asks you to trust

Custody asks you to trust solvency, honesty and operational security indefinitely. Unscoped API access asks you to trust that the key's permissions were configured correctly and stay that way. A scoped agent asks you to trust that the protocol enforces its own scope — a materially smaller and publicly verifiable assumption.

None of these models makes a strategy better. They determine whether a strategy's failure is the only way you can lose.

The honest case for custodial systems

Custodial products can offer things a scoped agent cannot: netted margin efficiency across strategies, execution from a single large account, and simpler onboarding for users who do not want to manage a wallet at all. Those are real advantages, paid for with counterparty risk.

HyperMirror takes the opposite trade deliberately — accepting fragmented margin across isolated sub-accounts in exchange for you never handing over control.

Neither model removes trading risk

Non-custodial architecture addresses who can take your funds. It says nothing about whether the positions taken on your behalf will be profitable, and perpetual futures remain capable of substantial loss in either model.

Past performance is not indicative of future results. Perpetual futures are leveraged instruments and carry a substantial risk of loss, including the loss of your entire position.