We Audited the Top of a Copy-Trading Leaderboard. Two of Three Failed.
Matrix CR Studio, Costa Rica. 2026-09-09.
Copy-trading runs on a simple promise: follow the traders at the top of the leaderboard and share in what they make. The leaderboard is the trust. So we tested it.
We took the highest-ranked traders on a public, fully on-chain perpetuals leaderboard, the same class of record a copy tool surfaces and a user follows, and we ran each one through an audit built to disprove it rather than confirm it. Everything below is from public on-chain data. No private information, and no cooperation from any platform.
## What we found
Two of the top three did not survive scrutiny.
Three records is a small sample, and this is not a claim about what fraction of traders fail in general. The point is narrower and more useful: the audit can tell a clean record from a dangerous one, and on this board it did.
Trader A looked like the strongest record on the board: a multi-million-dollar account and a double-digit monthly return, steady across the windows the leaderboard shows. Under scrutiny, the trader's net realized profit and loss was negative, and the risk-adjusted return (a standard measure of return against the risk taken) was below zero. The record that ranked highest had not made money on a realized, risk-adjusted basis.
Trader B posted the highest monthly return of the three, well ahead of the others. Under scrutiny, that return was built on a martingale: position size increased into losses. The result is a high return that sits one bad losing streak away from a severe drawdown.
Trader C passed. Positive risk-adjusted return, no increasing size into losses, low dependence on any single position, drawdown under control. A genuinely clean record.
The pass is the part that matters. The audit did not reject everyone. It passed the clean record and failed the two that only looked elite. It tells them apart, and that is the reason to trust the result: a test that fails no one is not a test.
Why a leaderboard can reward the mirage
Leaderboards rank on the numbers that are easy to show: return, equity, consistency. Those are exactly the numbers Trader A and Trader B look great on. The numbers that separate them from Trader C are harder: realized versus unrealized, risk-adjusted return, and whether the gains come from increasing size into losses. Those are not on the leaderboard. So a ranking can reward the record that is most dangerous to copy.
A headline profit is often mark-to-market on open positions. It is not money banked, and unrealized gains can disappear when positions resolve. A copier following the headline is following a number the trader's own settled history may not support.
What to check before you copy or allocate
- Realized, settled profit and loss, not the headline that includes open positions.
- Risk-adjusted return, not raw return.
- Whether the gains come from increasing size into losses.
- Enough settled history for the record to mean anything.
None of this requires the platform's cooperation when the trades are public. It requires the will to run the record through a test designed to fail it.
Why we built this to fail its own book
That is the work we do at Counterfoil: the receipt that a track record survived adversarial scrutiny, or the receipt that it did not. We built the method on our own trading first. When the live results came in below the paper ones and the book finished in the red, the same audit flunked our own record. An audit that will fail the hand that built it is the only kind worth trusting.
If you lead trust or risk at a platform that lists traders, or you allocate to one, the number at the top of your leaderboard is a claim. Name one record and we will return the receipt on it. We can tell you whether to believe it.