Why isn't the #1 trader on a leaderboard evidence of skill?
Because the top of a board is the maximum of many draws. Among 48,407 accounts, someone is on a streak by chance alone; the best of 48,407 has to clear t ≥ 4.90, not 2. On 2026-10-30, 0 of the 98 scoreable wallets in Hyperliquid’s all-time PnL top 100 did.
The bar rises with the number you picked from
| Picked from | t needed |
|---|---|
| 1 | 2.00 |
| 10 | 2.84 |
| 100 | 3.51 |
| 1,000 | 4.08 |
| 48,407 | 4.90 |
The bar keeps the chance that a no-skill account clears it, anywhere on the board, at the 2.3% a single test allows (Bonferroni on α = 1 − Φ(2)). Counting every account as independent overstates the charge when accounts copy each other; the true number of independent bets is unknown, so the table is an upper bound on the bar and the honest range sits below it.
Profit ranks are not return ranks
A PnL board ranks dollars. A large account with an ordinary return outranks a small one with an exceptional return, and PnL booked into a near-empty account cannot be turned into a return at all. In the all-time PnL top 10 on 2026-10-30, 1 wallets had most of their PnL booked while the account held under $1,000.
What would count
A record found after the fact has already been picked for looking good. The clean test is forward: freeze the list, then watch. We froze 200 wallets from the top of the board on 2026-10-14 and publish how each does from then on, on the wallet page.
Related questions
- Is this Hyperliquid trader skilled or just lucky?
- How many months of backtest or track record are enough?
- Why isn't the #1 trader on a leaderboard evidence of skill?
Numbers rebuild daily from latest.json. Method: Method. Not investment advice.