Why copying profitable Hyperliquid traders usually fails

The obvious plan is to find the most profitable traders and copy them. We tested it on Hyperliquid's own data, with real fees and real delays. It mostly doesn't work, and the reasons tell you what does.

What we found

• Past profit barely predicts next month. Of traders profitable over 90 days, about half were still profitable the following month. A coin flip.
• One trade makes the month. For a typical trader, about 90% of a month's profit comes from one trade. A copy that misses it, or catches it late, misses the month.
• Costs beat small moves. A typical copied trade moved less than its full round-trip cost of about 0.29% (fees in and out, slippage, and a worse entry than the trader's).
• Smooth curves hide open losses. Of 18 traders whose closed trades drew a smooth upward curve, 10 weren't smooth once open positions counted, and copies of them lost money.
• Our own score doesn't predict either. We rank traders, then check a month later. The top half does about as well as the bottom half, which is why we call the score a record, not a forecast.

What works better

• Select for copying, not for showing off. What matters is whether a copy of the trader would have made money: after a late entry, fees and slippage, over a recent window.
• Manage the entry. A copy doesn't have to enter the moment the trader does; how and when it joins decides whether the delay is a cost.
• Many small copies. Spreading across traders smooths the one-trade-a-month problem.
• Slower traders. The longer they hold, the less a delay costs.

Steady is built on these ideas, with its exact rules kept private. Its real-money results are on its page, after all fees.

Read the rankings for what they are

Our rankings describe how traders have done, open positions included. That's useful for understanding a trader, spotting luck and avoiding traps. It isn't a list of whom to copy. Copying can lose money.