Some wallets don't trade — they mirror someone else's trades with a bot. These are the groups we can prove, with what the copiers actually earned.
Across 105 clusters whose leader is in profit, leaders made $37.9M while the wallets copying them made $87.4K — 0.2% of the result they were following. The median copier captured 0.07% of their leader’s profit, and 37% of those copiers ended up down.
Same trades, opposite outcomes: entries get mirrored, but exits, position size and leverage do not. In 18 of 51 clusters with a profitable leader, the copiers still lost money. The other 64 tracked leaders are in the red — $156.0K lost between them, while their copiers finished at -$45.8K: the entry got copied, the result didn’t. This is our own data — an exchange sees its fills but not who is mirroring whom.
$325.5M is being copied in total, and it is not spread evenly: 81% of it follows just ten leaders, with half of it behind 3 of the 177 we track. A crowded leader is worth knowing about before you copy him — the same entry, shared by more size, fills worse for whoever arrives last.
The giveaway is the lag. When a dozen wallets enter the same coin seconds — in the tightest groups, a fraction of a second — after one trader, every single time, that isn't a hunch they share — it's one piece of software on one timer. We only count a wallet as a follower when the leader explains most of its trades, across several coins, at a steady delay.
What the numbers keep showing: copying doesn't carry the leader's result. The entry gets mirrored; the exit, the size and the leverage don't.
Copy trading means mirroring another trader's positions automatically: when they open a trade, the same trade opens on your account. On Hyperliquid every fill settles on-chain, so copying can be observed from public data — which is how the clusters above were found.
Usually not to the same degree. Across the clusters we track, followers consistently earn a fraction of what the leader earns, and some lose money on a leader who finished in profit. The entry gets mirrored — the exit, the position size and the leverage generally don't.
By the lag. A follower enters the same coin at a steady delay after the leader — seconds, in the tightest groups a fraction of a second — across many coins, over days. Human agreement is never that punctual — that pattern is software on a timer. A wallet counts as a follower only when the leader explains most of its trades.
Yes. Compass never holds your funds. Trades are mirrored from your own wallet, and you can revoke the delegation at any time.
Leaders have no control over who mirrors them — anyone can watch a public chain and follow it. That is a property of an on-chain exchange, not something Compass enables.
How long the record runs, how deep the drawdowns go, and how much leverage produced the result — not the headline profit. A large gain built on 40x leverage could as easily have gone the other way. The leaderboard shows those columns side by side.