Start with the time window and data coverage
A fair comparison of two Hyperliquid traders uses the same time window and the same data coverage for both: 24 hours, 7 days, 30 days or all time. Metrics taken from different windows, or from records that start at different dates, describe different things and should not be ranked against each other.
Hyperliquid publishes account value and PnL histories through its public info API, but fill queries are limited: a userFills request returns at most the 2,000 most recent fills, and time-range queries reach only the 10,000 most recent. Perpex therefore builds closed-position statistics from its own fill data set, which begins in March 2025, so an all-time trade count describes the covered period only. Check when a profile was last updated, because metrics are recalculated on a schedule.
Read PnL and ROI together
Period PnL is the dollar change in a wallet's results over a window, and ROI expresses that change as a percentage of the account value at the start of the window. Reading both shows whether a large profit came from a large account or from a high return on a small one.
Period PnL includes realized results and unrealized PnL on open positions, so it can move without a new trade. When the starting value is under $100, zero or unavailable, ROI is not measurable, and Perpex shows “—” instead of a percentage. For example, a wallet that starts a month with $1,000 and gains $500 shows 50% ROI, while a wallet that starts with $100,000 and gains $5,000 shows 5%. The first return is higher but was earned on far less capital.
Deposits and withdrawals during the window can distort ROI. Perpex trader profiles also show a time-weighted return, which links each period's gain or loss together so money moving in and out does not distort it.
Win rate, trade count and drawdown
Win rate is the share of closed positions that ended in profit, trade count is the number of closed positions, and max drawdown is the deepest fall in results from a previous peak. Together they show how often a trader wins, how large the sample is and how bad the worst losing stretch has been.
A position built and closed over several fills counts once, and a high win rate over a few trades is a small sample. On Perpex, max drawdown is the largest fall of the 30-day time-weighted return from a previous high, so deposits and withdrawals do not count as losses.
Read drawdown next to win rate. For example, nine winning trades of $100 and one losing trade of $1,000 give a 90% win rate and a net loss of $100.
Check leverage, concentration and open risk
Open risk is the loss a wallet could still take on the positions it holds now, and it depends on leverage, how concentrated those positions are and how close they sit to liquidation. A clean record of closed trades can sit next to a large open position whose loss has not been realized yet.
On Hyperliquid, the margin needed to open a position is position size times mark price divided by leverage. Liquidation begins when account equity falls below maintenance margin, which is half of the initial margin at max leverage. Cross margin is the default and shares collateral between all cross positions, so a loss in one can put the others at risk, while isolated margin keeps an asset's collateral to that asset.
Perpex trader profiles score exposure (open notional divided by account value), notional-weighted leverage, the largest position's share of open notional and the average distance from mark price to the reported liquidation price. Results that come from one market or one large position carry different risk from results spread across many trades.
Profit factor, average win and average loss
Profit factor is the total profit from winning closed trades divided by the total loss from losing ones, and average win and average loss show the typical size of each. A profit factor above 1 means winners earned more than losers lost, which reveals the pattern that a win rate alone can hide.
For example, a trader with 40 winners averaging $300 and 60 losers averaging $100 has a 40% win rate and a profit factor of 2.0, because $12,000 of gains is twice $6,000 of losses. A trader with an 80% win rate can still lose money if the profit factor is below 1. Perpex shows these figures before fees, so a very active wallet with a profit factor close to 1 may be unprofitable once trading fees are counted.
Compare one wallet with how most wallets fare
Most Hyperliquid wallets did not end with a realized profit. The Perpex trader report measured realized PnL across 751,941 Hyperliquid wallets from March 2025 to March 2026: 63% ended the period with a realized loss, 37% with a realized profit, and the median wallet's realized PnL was −$1.41.
The same report found realized gains highly concentrated: the top 0.1% of wallets accounted for $3.63 billion combined. Among trade-count groups, only wallets with 10K+ recorded trades had a positive median realized PnL, and 51% of them were profitable. These figures come from one dated snapshot (2026-03-20) of realized trading PnL, and whether fees and funding were deducted was not documented. Wallets are pseudonymous, so one person can control several wallets and a wallet can belong to a vault or a market maker. Treat a top wallet as one result from a large population and check whether its style and size are something you could follow.
Frequently asked questions
Why does a trader show “—” instead of an ROI?
ROI divides PnL by the account value at the start of the window. When that starting value is under $100, zero or unavailable, the percentage would be noise, so Perpex shows “—” instead. Rely on dollar PnL and the other metrics for that wallet.
Is max drawdown a percentage of the account?
Yes, in return terms. Perpex measures max drawdown on the 30-day time-weighted return, so 20% means the wallet gave back a fifth of its value since a previous high, with deposits and withdrawals left out. It is not a fall in cumulative realized PnL.
How many trades are enough to judge a win rate?
No number of trades makes a record reliable on its own, but more closed positions over a longer period give a sturdier figure. For reference, Perpex lists traders on the leaderboard only after at least 10 closed positions and ranks them by win rate only after at least 100.
Does strong past performance mean copying will be profitable?
No. Past performance does not predict future results. A copied account can also get different fills, sizes and costs than the source trader, so even a repeat of the trader's results would not carry over one-to-one.
What share of Hyperliquid wallets lose money?
In the Perpex trader report, 63% of 751,941 Hyperliquid wallets ended March 2025 to March 2026 with a realized loss and 37% with a realized profit. The figures come from one dated snapshot of realized trading PnL and do not predict any trader's results.