What happens to your positions when a leader is replaced

Replacement is an account event, not a scoring event

Scoring runs continuously. A leader's composite score drifts down over days or weeks as new closed trades arrive, and for most of that period nothing happens to your positions at all — a falling score reduces the weight the system wants to hold, but weight is a target, not an instruction to trade immediately. The separate note on how leader replacement works covers the rules that turn a score into a decision.

This note covers the other half: once the decision is made, what physically happens in your Hyperliquid account. That distinction matters because the two are often conflated. People assume a score drop means an immediate liquidation of their exposure, or conversely that a removed leader's positions linger indefinitely. Neither is true.

The short version is that replacement is a bounded, local operation. It touches one sub-account, it happens in a defined order, and it leaves the rest of the basket in exactly the state it was in before.

The timeline of a replacement

A handover proceeds in five stages. Each has a distinct effect on what you hold.

Stage 1 — flag and probation: no positions change

When a leader's score falls below the probation band, they are flagged. Existing mirrored positions in their sleeve are left exactly as they are, and the leader's subsequent entries and exits are still followed. What changes is capital flow: no new capital is routed into that sleeve, and if a rebalance runs during this period the sleeve's target weight is lower than it was.

Probation exists because score decay is noisy. A leader can have a bad fortnight inside a strategy that still works, and closing their book on the first weak signal converts noise into realised loss. The cost of waiting is that you continue to hold a deteriorating sleeve for a while. That is a deliberate trade-off, not an oversight.

Stage 2 — removal: new entries stop before anything is closed

Once removal is triggered, the first action is to stop mirroring new entries from that leader. This ordering is important and worth stating explicitly: the system does not close your position and then keep following the leader, and it does not flip the sleeve to the opposite side. It stops opening, then it starts closing.

In practice this means that if the removed leader adds to a position five minutes after the trigger, that addition is not mirrored. If they close part of a position, that close is mirrored, because reducing exposure in a sleeve that is being wound down is aligned with where the sleeve is going anyway.

Stage 3 — unwinding the open positions in the sleeve

The open perpetual positions in the removed leader's sub-account are closed at market. Whatever direction and size the sleeve held is reduced to zero, in that sub-account only.

Because the sleeve is isolated, this is a self-contained operation. If another leader in the basket is long the same market, their long is in a different sub-account and is not affected — it is not netted against the unwind, it is not used as an offset, and the unwind does not reduce the margin available to it. This is the concrete payoff of one sub-account per leader, and it is the reason a handover in one sleeve cannot disturb nine others.

Unwinding is where the honest caveat belongs. The closing fills you receive are market fills at the moment of the unwind. They are not the leader's fills, they are not a target price, and in a thin or fast market they can be worse than the mid-price you see on a chart afterwards. The larger the sleeve relative to book depth in that market, the more that matters.

• The unwind is scoped to one sub-account; other sleeves are untouched.
• Positions are closed at market, so execution quality depends on liquidity at that moment.
• No offsetting or netting against other leaders' positions occurs at any point.

Stage 4 — the sub-account is left flat

After the unwind the sub-account holds USDC margin and no positions. This is a real state you may observe for a period, especially after an emergency removal, and it is not an error. Capital sitting flat is producing nothing, which is a genuine cost, but it is preferable to holding exposure that the system no longer has a scored reason to hold.

If no replacement leader currently qualifies — for example if the candidate pool is thin, or you are in Starter Mode with a single sleeve — the sleeve can remain flat until one does. The system does not lower its selection bar to avoid an empty sleeve.

Stage 5 — redeployment to the incoming leader

The freed capital is allocated to the incoming leader at their target weight, which is derived from their composite score relative to the rest of the basket, not from whatever the outgoing leader happened to hold. A high-scoring replacement can receive more capital than the leader they replaced; a marginal one can receive less, with the remainder redistributed across existing sleeves at the next rebalance.

From that point the new sleeve behaves like any other: it mirrors the incoming leader's entries and exits, its PnL is measured separately, and its own score is recomputed as its trades close.

What you see in the dashboard at each stage

The handover is visible rather than silent. During probation the sleeve still appears in the following list with a reduced allocation. During the unwind, open positions for that sleeve shrink toward zero in the open positions section, and the closing fills appear in your mirrored trades feed with realised PnL attached. After the unwind, the sleeve shows allocated capital with no open positions. After redeployment, the leader address in the following list changes and new positions begin to appear.

If you are watching an unwind live, the thing to check is your own account totals rather than any single row: account value and unrealised PnL both move as unrealised becomes realised, and the two changes are the same event seen twice.

Realised versus unrealised PnL across the handover

This is the most common source of confusion during a replacement. Before the unwind, the sleeve's gain or loss is unrealised — it exists as a mark against open positions and moves with price. After the unwind, that same gain or loss is realised and no longer moves.

Nothing is destroyed or created at the moment of the close. If the sleeve was down 4% and it is closed, you did not lose 4% at the moment of closing; you had already been carrying it. What the close does is remove the possibility of that number recovering, which is precisely the point when the reason for holding the position no longer stands.

The practical consequence is that replacements make your realised PnL line jump in a way your account value does not. That is expected. Judge a handover by the change in total account value across the window, not by the realised figure in isolation.

The cost of a replacement

Handovers are not free, and it is better to state the cost than to imply that rule-based replacement is cost-free discipline.

• Exit turnover: closing the outgoing sleeve pays exchange fees and the builder fee on that mirrored volume.
• Entry turnover: opening the incoming leader's positions pays them again.
• Spread and slippage on both legs, worse in thin markets and during volatility.
• Funding paid while the sleeve is being wound down, if the position is on the paying side.
• Idle capital between the unwind and redeployment, which earns nothing.

Emergency removal: when the trigger is risk, not decay

Score decay is slow and gets probation. A hard risk breach does not. Liquidation-level exposure, an abrupt change in how a leader trades, or extended inactivity trigger immediate removal, and in that case stages one and two collapse into each other: new entries stop and the unwind begins without a probation period.

Emergency handovers are more likely to happen in exactly the market conditions where fills are worst, because those conditions are what produced the breach. This is an unavoidable feature of acting on risk rather than waiting. It is also the case that the sleeve is more likely to sit flat afterwards, since an emergency removal is not usually coordinated with a replacement candidate being ready.

What replacement does not do

Four boundaries are worth being explicit about, because each one is something users reasonably worry about.

• It does not net the outgoing sleeve against other leaders' positions. Isolation holds throughout the handover.
• It does not move, withdraw or hold your funds. The agent permission is trade-only; capital stays in your own Hyperliquid account.
• It does not touch positions you opened yourself outside the mirrored sleeves.
• It does not require you to sign anything. Replacement is execution within the permission you already granted, not a new approval.

Conclusion: the point of isolation is that a handover stays local

A leader replacement is a contained sequence: stop entries, close that sleeve at market, leave it flat, redeploy at the new target weight. It realises PnL that already existed, it pays turnover twice, and it can be executed at a poor moment when the trigger is risk rather than decay.

What it cannot do is spread. In a single-account copy setup, closing one strategy's book affects the netted position, the margin and therefore every other strategy in the account. With one sub-account per leader, the handover ends at the boundary of that sleeve — which is the whole reason the isolation exists.

If you want to watch a handover rather than read about one, the dashboard shows each sleeve, its open positions and its current allocation as they change, and the live model portfolio on the homepage shows the current basket before you connect anything.