HyperMirror vs Hummingbot

Quick verdict

HyperMirror fits when you want exposure to skilled discretionary traders without building or running anything.

Hummingbot may fit better when you want to run your own quantitative strategies and have the engineering capacity to operate them.

Custody and security

Self-hosting means nobody stands between you and the venue, which is the strongest possible position on operator risk and the weakest on operational risk. Keys sit in your environment, and the practical threat is mundane: a leaked credential, an exposed dashboard, a stale key on a machine you stopped patching.

HyperMirror substitutes a narrow trust boundary for that responsibility. The Hyperliquid agent can sign trade actions and nothing else, so even total operator compromise cannot produce a withdrawal; funds stay in your own account and the approval is revocable from your wallet. That is a real trade — operator trust in exchange for not being the on-call engineer.

• Self-hosted: zero operator trust, complete operational responsibility.
• Hosted agent: trade-only permission with no withdrawal path.
• Neither approach lets anyone move funds out of your Hyperliquid account.
• Market risk is identical in both cases.

Fees and incentive alignment

Open-source software costs nothing to license. Market making in particular has a different cost structure anyway: your economics are dominated by fee tiers, rebates, latency and inventory management, not by software licensing.

HyperMirror is priced as a single volume-linked fee — 0.1% of mirrored notional through Hyperliquid's builder-fee mechanism, no subscription, no profit share. For a high-frequency market-making strategy that fee model would be entirely inappropriate, which is a fair illustration of how differently the two products are aimed.

How copying and automation actually work

Market making earns spread by continuously quoting both sides and managing inventory; arbitrage earns from price differences across venues. Both are latency- and infrastructure-sensitive, both require careful risk parameterisation, and both can be profitable in market conditions where discretionary directional trading is not.

Copy trading has none of that machinery. It identifies traders whose public Hyperliquid history shows consistency, discipline and survivability, sizes them by composite score, and reproduces their positions in your own account with one isolated sub-account per leader. The skill being rented is human judgement about direction and timing, not microstructure.

• Different edge: human discretionary skill, not spread capture.
• Latency-tolerant by nature; not a high-frequency system.
• Up to 10 leaders in Full mode; 1 until $100k mirrored volume.
• Isolated sub-accounts prevent cross-leader netting.

Risk controls and leader failure

A self-hosted market maker's risks are inventory and operations: quoting into a one-way move, a process failure leaving orders live, a connectivity issue at the worst moment. Mitigating them means engineering — kill switches, inventory limits, alerting — and being available when they fire.

A copy basket's risks are leader-shaped: edge decay, discipline breakdown, correlated positioning. Governance covers the first two with soft-issue strikes, replacement, and immediate emergency removal on defined rules rather than on relative score, and closes the departing leader's sub-account positions. Correlation across leaders remains a genuine exposure that diversification reduces but does not remove.

Who should choose which

Choose HyperMirror if: you want a maintained allocation with no infrastructure, you are comfortable with a trade-only agent as the trust boundary, and your interest is Hyperliquid directional exposure.

Choose Hummingbot if: you want to run market-making or arbitrage strategies, you have the engineering capacity to operate them, or you want fully inspectable, modifiable software.

Limitations, on both sides

Where HyperMirror is limited: no market making, arbitrage or custom strategy logic; Hyperliquid only; closed allocation logic, though every action is verifiable on-chain; and no ability to pick leaders yourself.

Where Hummingbot is limited: significant engineering and operational burden; strategy performance depends on infrastructure quality and fee tiers; key management is entirely your exposure; and there is no leader-selection or copy-trading layer at all.

Methodology, in one paragraph

HyperMirror's side of every row above follows one published rule set. The Elite basket is sticky: a leader stays mirrored until a rule removes them, and a higher-scoring wallet elsewhere never forces a swap. Emergencies — account value below roughly $1,000, or no fill for 96 hours or more combined with zero trades in 7 days — remove a leader immediately. Everything softer accrues at most one strike per wallet per UTC calendar day, with three strike-days triggering replacement and a clean day resetting the counter; drawdown for that test is measured on jump-adjusted equity so deposits and withdrawals are not read as trading losses, and ROI is measured from PnL. The full table is on How it works and in the documentation.

Risk statement

Structure changes which risks you carry, not whether you carry risk. Nothing on this page is a return estimate, a recommendation, or a claim about another provider's results — check any provider's own documentation before deciding.

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.