The first week is where most of the mistakes happen, and almost all of them come from reading a small sample as if it were evidence. This page states what normal looks like, what is worth watching, and which reactions to resist.
In short
In the first week, watch four things: whether each sleeve is trading at all, margin used against account value, funding accruals on held positions, and any replacement events. Expect flat days, one sleeve red while the basket is up, and entry prices that differ from the leader's. A week is far too short a window to judge a basket on.
At a glance
First-week observations: normal, expected, or worth checking.
Observation
Reading
Action
A sleeve has no position for days
ReadingNormal — that leader has not traded
ActionNone
One sleeve red, basket up
ReadingNormal — dispersion is the mechanism
ActionNone
Your entry differs from the leader's
ReadingExpected — mirrored orders are new orders
ActionNone
Funding debited on a held position
ReadingExpected — accrues to the holder at the stamp
ActionNone
A leader is replaced
ReadingPolicy working — emergency or three strike-days
ActionRead the replacement page
No activity anywhere while leaders trade
ReadingWorth checking
ActionVerify registration and readiness
Positions larger than you want
ReadingSizing decision, not a fault
ActionReduce capital in the account
What a normal week looks like
A score-weighted basket does not produce a smooth line. On any given day some sleeves are flat because their leaders did not trade, one or two are red, and the aggregate can be either. That dispersion is the mechanism working, not a malfunction — the point of holding several independent leaders is that they do not move together.
Flat is also a legitimate state. Leaders with slower strategies can go days without a fill, and a sleeve with no position is not a sleeve that has failed.
Some sleeves flat, some open, some red — simultaneously, on the same day.
Entry and exit prices that differ from the leader's on every trade.
Funding paid or received on positions held across a funding stamp.
In Starter mode, a single sleeve doing all of the work.
The four numbers worth checking
Per-sleeve activity tells you whether the basket is actually engaged. Margin used against account value tells you how much of your capital is currently at work. Funding tells you what holding costs you regardless of direction. Replacement events tell you whether the basket composition changed under you.
Everything else — an individual trade's PnL, a single sleeve's day, a percentage move on the account — is noise at this horizon.
Isolation is what makes the week readable
Because each leader runs in one Hyperliquid sub-account per mirrored leader, you can attribute results per leader instead of staring at one netted position and guessing. A bad week is traceable to a specific sleeve, and a good one is not credited to the wrong leader.
That attribution is the practical benefit of isolation in week one. It also means a liquidation in one sleeve consumes that sleeve's margin and stops there rather than reaching other leaders' collateral.
The three reactions that usually hurt
First, pulling capital after one red sleeve. Isolation exists so that one leader's bad run is one sleeve's bad run; exiting the whole book converts a bounded loss into a realised one across the basket.
Second, judging the basket by its worst member. A weighted set of independent leaders is designed so that no single one determines the outcome, which necessarily means at least one member usually looks bad.
Third, judging anything at all on a few days. Selection is built on multi-month evidence and replacement is deliberately slow on soft signals precisely because short windows mislead.
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.
When something is genuinely worth acting on
Two categories justify action. Operational: no sleeve has traded at all for an extended period while the leaderboard shows active leaders, or the dashboard does not reflect your account. Personal: the position sizes are larger than you are willing to hold, in which case reduce capital rather than waiting for a drawdown to make the decision for you.
Stopping autopilot and revoking the agent approval are separate actions, and both remain available at any time without funds moving anywhere.
Methodology
Scoring and replacement are documented in full on How it works and in the Docs (Policy v3). In short: the Elite basket is sticky, emergencies remove a leader immediately, and soft issues accrue at most one strike per UTC day with three strike-days triggering replacement. Read how it works or the documentation for the full table.
Questions
Frequently asked
One of my leaders is down. Should I stop?
A single sleeve being down is the expected behaviour of a diversified basket, not a signal. Isolation already bounds that sleeve's loss to its own sub-account.
Is a week enough to judge performance?
No. Selection is based on multi-month evidence, and a week of results is dominated by noise in either direction. Past performance is not indicative of future results.
Why do my prices differ from the leader's?
A mirrored order is a new order placed after the leader's fill is already public, so latency, book depth and rounding all move your average price. The tracking-error page covers each driver.
What happens if a leader is replaced during my first week?
The replaced leader's sleeve is wound down and a new leader takes its allocation. Membership changes only on an emergency trigger or three strike-days, never on rank alone.
