The leaderboard tells you who made money. It does not tell you whether copying them would have made you money. A trader who made $10M scalping illiquid coins at 40× cannot be copied profitably by anyone; a trader who made $400k holding ETH for two days at 3× can be copied by almost anyone. Here is how to tell them apart from the numbers hypercopy shows.
start with the filters hypercopy already applies
Before you see a name, the bot has removed:
- accounts under $10,000;
- market makers: accounts whose volume divided by account value divided by days is above 20. They trade thousands of times a day for a spread you cannot capture;
- vaults and agents: their fills are pools or bots, not a person;
- fast limit-order traders: a trader whose volume is mostly resting (limit) orders is kept only if they hold their positions an hour or more and make at most 100 fills a day. Faster than that, it is a market maker's pace, and a copy of it is a taker order chasing a resting order that just got hit;
- fewer than 30 fills, fewer than 5 round trips, or a median hold under five minutes;
- traders with less than 80% of their volume on core perps.
What is left are humans trading hyperliquid perps at a pace and size that can be followed. The list shows them sorted by return per day held (what each dollar in their trades made per day), with only those who have 10+ trades, a positive 30-day ROI and a positive all-time PnL. Filters narrow it by profile (whales, scalpers, swing…) and by effective leverage (≤ 5×, 5–10×, > 10×). Now the profile.
pnl or roi, and which window
PnL is dollars made. ROI is PnL relative to account size, and hypercopy's ROI excludes deposits so a whale topping up does not look like a loss. Big PnL with small ROI means a big account taking small risk: usually easy to copy. Big ROI with small PnL means a small account taking big risk: often a lucky month.
Windows: 24h is noise. 7d shows the current streak. 30d is the default for a reason: long enough to include a bad week, short enough to describe the trader as they are today. All time tells you they survived, which matters, but it can hide a change of style.
Look at a trader in two windows. Someone up 30d and up all time is consistent. Someone up 30d and flat all time just had a good month.
median hold
The single most useful number. It is how long the trader's typical position stays open.
- Minutes: a scalper. Their edge is in being first. You are never first. Skip, however good the PnL.
- Hours: a day trader. Copyable on liquid pairs if you keep your share small, because slippage is a real share of their profit.
- Days: a swing trader. The easiest to copy: a few basis points of slippage on a trade that makes 5% is noise.
trades per day
How much of your attention (and fees) the copy will take. Twenty trades a day at 0.05% builder fee plus 0.045% exchange fee each way is about 4% of the copied notional per day in fees. Two trades a day is 0.4%. A trader whose edge is smaller than that cannot be copied profitably, whatever their PnL says.
average trade size
It decides how much you need. In pro rata, the bot turns it into an advised minimum amount: 10 × their capital ÷ their average trade, at least $20. Under it, most of their positions would be under $10, hyperliquid's minimum, on your side. With a % per trade, multiply it by your share: under $10 the copy is skipped, and over your max per trade you take less of their big trades than of their small ones.
max drawdown (30d)
The worst peak-to-trough fall of their account over the last 30 days. A trader with 40% ROI and 35% drawdown spent part of the month down a third; copying them means you would have too. Decide what you can sit through before you look at returns, then filter by that.
open positions and leverage
The profile shows the leverage they really put to work: usual, and at their peaks over 30 days, as their exposure over their account. A pro rata copy uses that real leverage (capped by the market and by your own cap, if you set one); a fixed $ or % copy uses theirs up to your maximum (10× by default). 3× on BTC and 20× on a mid-cap are very different liquidation distances. See leverage and liquidation.
a checklist
- 30d ROI positive, all-time ROI positive.
- Median hold of hours or more.
- Trades per day in the single digits.
- Max drawdown you would accept on your own account.
- An amount at least the advised minimum the bot shows for them (with a %: average trade size × your share ≥ $10).
- Leverage on their open positions that you could stomach.
Then copy small, and give it two weeks before you scale. The trader you copy will have a bad week; the question is whether you sized for it.
faq
should i copy the number one on the leaderboard?
Not blindly. The list is sorted by return per day held, which rewards efficient trades, not size; still open the profile and check the drawdown, the hold and the leverage before you copy.
should i copy several traders?
Yes, if they trade differently. Two swing traders on BTC and ETH will be long at the same time; a copy of both doubles the same bet. Up to ten copies per account.
can i see a trader's win rate?
Yes: the profile shows their winning trades, long and short. It is a poor guide on its own, though: a trader who wins 40% of trades with 3:1 payoffs beats one who wins 70% with 1:2.
copy hyperliquid's best, from telegram.
pick a trader, set your size. hypercopy trades with a key that can never withdraw, and you can stop at any time.
start copying