A trader's return is not what you would have made copying them. A copy is sized to your account instead of theirs, pays a fee on every order, skips any order too small for Hyperliquid to accept, and can't follow a trader into markets Hypertide doesn't support. This page runs that arithmetic over every trader on our board who has enough history to measure, using their settled on-chain fills.
Data as of 2026-09-10 · recomputed hourly from the latest leaderboard refresh
In every window with enough traders to report, some who were profitable for themselves would not have been profitable for a $10,000 copier.
| Window | Tracked | Enough history | Profitable for trader | Not profitable for a $10,000 copier | Median trader | Median copier |
|---|---|---|---|---|---|---|
| 1d | 176 | 21 28 excluded: over 2,000 fills | 15 | 13 of 15 6 at 0.0% | +8.0% | 0.0% |
| 7d | 176 | 11 7 excluded: over 2,000 fills | 8 | 2 of 8 1 at 0.0% | +8.9% | +0.2% |
| 30d | 176 | 1 7 excluded: over 2,000 fills | Too few to report — fewer than 3 traders cleared this window's bar with a complete history. 7 more had over 2,000 fills in the window and can't be measured across all of it — see below. | |||
| 90d | 176 | 2 4 excluded: over 2,000 fills | Too few to report — fewer than 3 traders cleared this window's bar with a complete history. 4 more had over 2,000 fills in the window and can't be measured across all of it — see below. | |||
Read the counts, not the percentages. The long windows hold single-digit samples.
Tracked is every trader on the board. Enough history is how many cleared that window's bar — only those are measured, and every other number in the row is out of them. The bars, which scale with the window:
Plus the recency and drawdown rules on the TrustScore page. A trader who does not clear them is not counted as a success or a failure — they are not counted.
The 1d row has the largest sample and the least meaning: one day says little about what copying someone is worth. It is shown because leaving out the biggest sample would be its own kind of cherry-pick.
A copier return of 0.0% is not a loss. It means almost nothing was copied. When we re-ran the simulation for individual traders we found two causes: the trader trades only markets Hypertide can't place an order on today — Hyperliquid's builder-deployed (HIP-3) markets — or the copies came out smaller than Hyperliquid's minimum order and were skipped. Those copiers made nothing, so the table counts them as not profitable, and shows how many there are so they aren't mistaken for losses.
Hyperliquid returns a trader's most recent 2,000 fills per request. When a trader has more fills than that inside a window, we can see their return over the whole window but can only simulate a copier over the most recent part of it. Comparing the two would set a week of copying against a month of trading, so those traders are left out of the comparison rather than guessed at. The table still counts them, under "Enough history", so you can see how many.
The most active traders are the ones most likely to be left out, which is why the long windows are thin.
The simulated copier fills at exactly the trader's own price, with no delay. A real copy is placed after the trader's fill arrives and may fill up to 0.5% worse. Each of those real-world costs makes a copier's result worse, never better — so the true gap is at least as large as the table shows.
The copier is a $10,000 account, net of copy sizing, fees, and orders below Hyperliquid's minimum that a copy would have skipped.
It does not say copiers lose money. In some windows the median copier return is positive. It says the trader's number is not your number, and that the difference is large enough to change which traders are worth copying.
It is not a prediction and not financial advice. Past performance does not guarantee future results.
See which traders clear the bar