How to copy trade on Hyperliquid in 2026

Before you start — the prerequisites

Four things need to be true before any of the steps below make sense. Getting them in place first removes most of the friction people hit halfway through a setup flow.

• A self-custody wallet you control, with the seed phrase backed up offline.
• USDC bridged or deposited to Hyperliquid, and sitting in the perpetuals context.
• An amount you are willing to see in a full drawdown, sized so a total sleeve loss is tolerable.
• A written exit condition — what would make you stop — decided before you start, not during a drawdown.

Step 1 — Fund your own Hyperliquid account

Everything downstream assumes capital sitting in a Hyperliquid perpetuals account you control. Bridge or deposit USDC to Hyperliquid, then make sure the balance is in the perps context rather than sitting on spot or unified — margin for perpetual positions has to be where the positions are, and this is the single most common setup blocker we see.

Do not size this as a test if you do not intend it to be one. Copy systems allocate proportionally to the capital they are given, and an account that is too small to express a leader's positions cleanly will track them poorly for arithmetic reasons rather than strategy reasons. The capital page covers where those thresholds fall.

Step 2 — Find candidate traders

Hyperliquid's public data means you can enumerate accounts and read their real history. The obvious starting point is the exchange leaderboard, and the obvious mistake is treating the leaderboard as a shortlist. Leaderboards rank on short-window PnL or ROI, which selects for leverage, small-account percentage effects and survivorship — the accounts that took the same risk and lost are not shown.

Better sources of candidates: accounts with long continuous histories rather than recent spikes, accounts whose size has grown through realized PnL rather than deposits, and accounts that trade the same markets repeatedly enough that their behaviour is characterisable. Breadth matters at this stage; you are building a pool to filter, not picking a winner.

• Prefer long histories over large recent numbers.
• Distinguish equity growth from deposits before reading any ROI figure.
• Collect a pool of candidates; do not shortlist from the top of a 7-day board.

Step 3 — Evaluate them honestly

The evaluation that matters is whether the record is repeatable, not whether it is impressive. Five properties do most of the work: consistency of realized PnL across periods, win rate read together with profit factor, position discipline (does size scale with conviction or with recent losses?), drawdown behaviour, and account survivability — whether the account has ever been liquidated or run margin close to the edge.

Two traps are worth naming. First, ROI without a known capital base is uninterpretable: percentage returns on an account that is being topped up or drawn down mean different things. Second, a high win rate alone is compatible with a strategy that takes small wins and occasional catastrophic losses; profit factor is what exposes that.

HyperMirror automates this filter, scoring candidates on those five inputs and requiring every floor to be cleared before a trader enters the basket at all. If you do it manually, apply floors before ranking, for the same reason.

Step 4 — Decide: single leader or diversified basket

Copying one trader is copying one strategy, one regime preference and one risk appetite. When that edge decays — and edges decay — the whole account decays with it, at full size, with no internal offset. It is also the version that feels best while it works, which is why it remains popular.

A basket spreads the dependence across several leaders, which lowers the variance contributed by any single one. It costs more in complexity and requires isolation to work at all, and it does not protect you from a market-wide move that hits every leader simultaneously. Diversification across correlated perp traders is real but partial; expect it to soften idiosyncratic failure, not systemic risk.

There is also a capital dimension: a basket only functions if each sleeve is large enough to express positions after order-size rounding. Below that, you are better off with fewer leaders sized properly than with many sized into noise.

Step 5 — Set up non-custodial mirroring

The non-custodial setup on Hyperliquid is three signatures and no transfers. Connect the wallet that owns the account. Approve a trade-only agent, which authorizes a named agent address to submit trading actions while leaving withdrawals signable only by you. Approve a maximum builder-fee rate, which bounds what can be charged on your fills.

At no point should you be asked to deposit into an operator's address, hand over a private key, or grant withdrawal rights. If any of those appear in a setup flow, the product is not non-custodial regardless of how it is described.

Once approved, mirroring runs against your capital in place. With HyperMirror each leader is given its own sub-account, so their positions never net against each other and each sleeve's results stay attributable.

• Connect wallet — proves ownership, moves nothing.
• Agent approval — trade-only permission, revocable on-chain.
• Builder-fee approval — caps the fee rate you can be charged.

Step 6 — Choose risk settings deliberately

The leader's risk appetite is not automatically yours. Two limits do most of the protective work: a per-leader notional ceiling, so no single sleeve can dominate the book, and a leverage cap applied independently of whatever leverage the leader chooses to run.

Think about the limits in terms of the worst case they permit rather than the returns they allow. If a sleeve were fully lost, what fraction of the account is that? If the answer makes you uncomfortable, the ceiling is too high — and no amount of leader quality changes that arithmetic.

Also decide in advance what would make you stop. Systems that get switched off in the middle of a drawdown, at the point of maximum discomfort, tend to lock in the drawdown and miss the recovery. Writing the exit condition down before you start is cheap; deciding it under stress is not.

Step 7 — Monitor what actually matters

Day-to-day PnL is the least informative thing on the screen. What is worth watching: per-leader attribution (is one sleeve carrying or dragging everything?), drift between intended and actual weights, tracking error against the leader, and whether any leader has accumulated a strike or been replaced.

Expect your results to differ from the published leader history. Your entries fill at different prices, your account size differs, and funding accrues at different points in your holding period. That divergence is structural, not a defect, and it goes in both directions.

What the first two weeks actually look like

The first fortnight is where expectations get recalibrated, so it helps to know the shape of it in advance. Days one to three are usually quiet: mirroring only acts when a leader acts, and a leader between setups produces no fills at all. An empty positions table on day one is normal, not a fault.

Once fills begin, the first thing you will notice is divergence from the published leader history — different prices, different size, funding accruing at different points. Judge the system on whether positions are being reproduced faithfully and in the right direction, not on whether the PnL matches. Two weeks is far too short a window to evaluate an edge; it is long enough to verify that plumbing, isolation and attribution are working.

• Day 1–3: verify the agent shows as approved and each sub-account is funded and reachable.
• First fills: check direction, market and rough size against the leader, not the PnL.
• Week 1: confirm per-leader attribution is separable — no positions netting across leaders.
• Week 2: review tracking error and weight drift; resist judging the edge this early.

Common mistakes

Most of the damage in copy trading comes from a short list of avoidable errors rather than from exotic risks.

• Copying the top of a short-window leaderboard, which selects for leverage and luck.
• Mirroring multiple leaders into one account, where opposing positions net out.
• Sizing the account too small for the number of leaders, so positions round into noise.
• Treating estimated model performance as a forecast of your own returns.
• Turning the system off mid-drawdown and back on after the recovery.
• Granting a permission broader than trade-only because a product asked for it.

Capital considerations

Copy trading has fixed frictions that do not scale down: minimum order sizes, rounding, and margin that has to sit idle in each sub-account. Below a certain account size those frictions dominate, and adding leaders makes it worse rather than better.

HyperMirror handles this with two modes. Starter mode mirrors a single leader with the same risk controls and isolation. Full mode, which unlocks automatically at $100,000 of mirrored volume, mirrors up to 10 scored leaders in parallel. The capital page covers where the practical thresholds fall in more detail.

Nothing here is financial advice. Perpetual futures are leveraged instruments: a position can be liquidated in full, and past performance of any trader is not indicative of future results. Copy trading does not remove that risk — it changes who makes the decision, not what the market can do to it.