Once you copy more than one trader, the interesting question stops being who to copy and becomes how much to give each of them. Equal weighting is the default because it requires no opinion. It is also the choice that treats your weakest qualifier exactly like your strongest one.
In short
Equal weighting gives every leader the same capital regardless of evidence quality. Score weighting allocates in proportion to a composite score built from realized PnL consistency, win rate, profit factor, position discipline and account survivability, so exposure scales with the strength of the evidence rather than with basket size.
At a glance
Equal weighting vs score weighting
Dimension
Equal weight
Score weight
Allocation rule
Equal weightSame capital to every qualified leader
Score weightCapital proportional to composite score
Treatment of a marginal qualifier
Equal weightIdentical to the strongest leader
Score weightSmallest slice in the basket
Response to decay
Equal weightNone until removal
Score weightWeight follows score at each rebalance, but the basket is sticky — a lower score alone never forces removal
Failure mode
Equal weightWeakest leader drags the whole basket
Score weightScore lags a genuine regime change
What equal weighting actually assumes
An equal-weight basket of ten leaders implicitly asserts that you have no ability to distinguish between them. If that were true, equal weighting would be correct — it is the right answer under total ignorance, and it is why index construction often starts there.
But a copy-trading basket is not drawn at random. Every member has already passed a selection filter, which means you do have information about them: how long their record is, how consistent it is, how much of it came from a handful of outsized trades. Discarding that information at the allocation step wastes the only work that distinguishes one system from another.
Equal weight maximises exposure to the marginal qualifier — the trader who barely cleared the floor.
It ignores record length, so a 90-day sample and a 900-day sample are treated identically.
It cannot express partial conviction: a leader is either fully in or fully out.
How composite scoring produces a weight
HyperMirror scores each candidate on realized PnL consistency, win rate, profit factor, position discipline and account survivability. Each dimension has a floor; failing any floor removes the trader from consideration entirely rather than reducing their weight. Only survivors are scored, and capital is then distributed in proportion to that composite score.
The practical effect is monotonic but bounded: a leader with twice the score does not receive twice the risk in an unbounded way, because per-leader notional ceilings cap any single allocation. Weighting decides relative conviction; the ceilings decide absolute exposure.
The failure modes of score weighting
Score weighting is not free of assumptions. It assumes the scored dimensions remain predictive, that the sample behind them is long enough to be meaningful, and that the market regime the score was earned in has not disappeared. All three can break.
The mitigations are structural rather than clever: floors that reject short or thin records, continuous rescoring so decay shows up quickly, emergency removal and soft-issue-strike replacement when it does, and caps that limit what any single mistaken weight can cost. None of that removes the possibility of loss.
Scores are backward-looking; a regime change can invalidate a high score without warning.
Concentration risk rises as one leader's score pulls away from the field — caps exist for exactly this.
Frequent reweighting increases turnover, and turnover costs money in fees and slippage.
Concentration is what ceilings are for
Score-proportional allocation is unbounded on its own. If one leader's composite score pulls away from the field, proportionality hands them a share of the book large enough to defeat the reason for holding a basket at all — and it does so quietly, as a by-product of arithmetic rather than as a decision anyone made.
Per-leader notional ceilings cap that. The division of labour is worth stating precisely: weighting decides relative conviction between qualified leaders, ceilings decide absolute exposure to any one of them. A ceiling is the layer that bounds what a single mistaken high score can cost.
Reweighting is not free
Every change in weights closes positions and opens others, and both sides pay taker fees, spread and slippage. That cost is certain; the improvement from the new weights is not. Any allocation policy that reacts quickly to score movement therefore transfers a predictable amount of capital to trading costs in exchange for an unpredictable benefit.
This is why weights are applied at rebalance rather than continuously, and why the policy is deliberately patient on ordinary variance while remaining immediate on a hard breach. Patience on noise and decisiveness on rules is a cost decision as much as a risk decision.
Turnover cost scales with how often weights move, not with how much they improve.
Continuous reweighting would pay fees to chase score noise.
A breach bypasses the schedule because the signal is not statistical.
What this means for your account
You do not set weights manually. The basket and its weights are published on the leaderboard, and the model portfolio shows what that allocation produced over recent windows. Both are estimates derived from public trader history, not a record of your account.
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.
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
Can I override the weights?
No. Weights are produced by the scoring model so that allocation stays systematic rather than discretionary. You control whether autopilot runs at all and how much capital sits in the account.
How often do weights change?
Scores refresh continuously against cached snapshots, and weights are applied at rebalance. An emergency trigger — such as account value falling below roughly $1,000 — removes a leader immediately rather than waiting for the next cycle. A higher score elsewhere never forces a replacement on its own.
Does a higher score mean higher expected return?
No. A higher score means stronger and more consistent historical evidence under the scored dimensions. It is a statement about the record, not a forecast.
