PnL vs ROI: Reading the Hyperliquid Leaderboard

A public perp leaderboard gives you two headline numbers per trader: how many dollars they made and what percentage return that represents. They sound interchangeable. They are not. Read together, they tell you something the dollar figure alone hides — how much capital a trader is risking to produce a result, and therefore how that result might behave next week.

Today's Hyperliquid leaderboard, ranked by 7-day realized profit, is a clean teaching case. Before the table, two definitions, because the whole point turns on them.

PnL (profit and loss) is the absolute dollar outcome over a window — here, the last seven days, realized. Realized means closed: positions that have actually been booked, not open marks that could still reverse. ROI (return on investment) is that same profit expressed as a percentage of the capital behind it. One is a dollar count. The other is an efficiency ratio. The bridge between them is account equity:

PnL ≈ equity × ROI

That single relationship is the lens. Hold it up to the leaderboard and the rankings reorganize themselves.

The leaderboard

The ranking is by PnL, top to bottom. Notice that ROI does almost the opposite. The trader at the top earned the most dollars on the lowest return. The traders near the bottom earned a fraction of the dollars on returns roughly ten times higher.

A worked example

Take the top line. An account with $4.56B in equity posted $102.7M of realized profit — a 2.3% return for the week. Run the bridge backwards: 2.3% of $4.56B is about $105M, which lines up with the booked figure. Nothing surprising there; that is just the identity confirming itself. What matters is the shape of it. This trader did not need an aggressive return to top the board. Size did the work. A 2.3% move on four and a half billion dollars dwarfs a 24% move on forty million.

Now the bottom of the table. 0xf822…e01a turned $38.4M of equity into $8.39M — a 23.9% return in seven days. That is an enormous weekly number on capital, and it is the highest ROI on the board, yet it ranks fifth by dollars because the base is small. 0x0ddf…a902 and 0xd475…1a91 sit in the same regime: equity in the tens of millions, ROI north of 22%.

So the same leaderboard holds two different activities. At the top, very large books extracting thin percentage returns at scale. Near the bottom, much smaller books extracting fat percentage returns — almost certainly through more concentrated directional exposure or higher effective leverage relative to capital. Same league table, opposite risk postures.

Why this matters for reading risk

ROI is the closest thing the board gives you to a risk signal, and the spread here is the story. When the largest accounts are content with low single-digit weekly returns, the marginal dollar of aggressive perp risk is not coming from them — it is coming from the smaller, high-ROI cohort. That cohort is where leverage and concentration concentrate. It is also where the variance lives: returns of 22–24% in a week are not a baseline you extrapolate, in either direction.

There is a survivorship trap baked into any leaderboard. You are looking at the accounts that won this week. The same aggressive posture that produced a 23.9% week is the posture that produces a deeply negative one when the tape turns, and those accounts simply drop off the board rather than appearing with a minus sign. High ROI is evidence of high sensitivity to the market, not evidence of skill that persists.

How to actually use the two numbers

When you scan a board like this, read the pair, never the dollar figure alone:

• High PnL, low ROI, large equity → scale, not aggression. The result is robust to small adverse moves and tells you little about directional conviction.
• Modest PnL, high ROI, small equity → aggression relative to capital. The result is sensitive, high-variance, and informative about where leverage is being applied — and where it can unwind.
• The gap between a trader's ROI and the board's median ROI is a rough read on how far out on the risk curve they are sitting this week.

None of this requires copying anyone. It is a way to infer the risk distribution of a venue from public numbers: this week, the dollar leaders are playing size, and the percentage leaders are playing leverage.

See the full breakdown and track it live at app.formion.ai.