What we measure on a vault, and what it actually means

A leaderboard is only as honest as the columns it shows you. The board has a dozen of them, and each one is there because it answers a question the headline APR doesn't. This is the field guide: what every column means, in plain English, and the trap each one is built to catch.

Hover any header on the live board and you'll get a one-line version of what's below. This is the long version.

The money columns

TVL — how much real capital is in the vault. Not the headline follower count (Hyperliquid's public list caps at 100, so it understates), but the actual account value. It's the one number that can't be flattered: a vault either holds real money or it doesn't. The trap it catches is the dust vault — a great-looking return on $300 that no serious money has ever trusted.

Age — how long the vault has existed. Boring, until you remember that two out of three Hyperliquid vaults are already dead. A vault that has survived a year through real volatility has shown you something a six-week-old can't. Age is the cheapest survivorship filter there is.

The return columns

Annualized return — but rebuilt from the P&L curve, not the equity curve. This is the single most important methodological choice on the board. Deposits make a vault's equity jump without anyone earning a cent; read the equity line naively and you'll credit the manager for other people's money walking in. I work from realized profit and loss instead. The whole confusion gets its own post — why "up" isn't the same as "made money".

Max drawdown — the worst peak-to-trough fall along the way. An APR tells you where the vault ended up. Drawdown tells you what it put you through to get there. A 60% drawdown means that at some point you'd have watched more than half your money disappear — and most people sell exactly there, at the bottom, which is the most expensive mistake in investing. When you see a drawdown above 100% on the board, that's not a metric anymore — it's a near-zeroing, a vault that effectively blew up.

Calmar and Sharpe — return per unit of pain. A raw percentage with no risk attached is a half-truth; Прибутковість з урахуванням ризику are the honest version. Calmar is annual return divided by the worst drawdown — my favorite, because it asks "how much did you make for the worst moment you caused?" Sharpe rewards a smooth ride over a jagged one. The trap both catch: a huge Calmar usually means a tiny base, not genius, which is why the board flags the extreme ones.

Consistency — how often the curve goes up. The share of periods that were positive. A vault can hit a target return through one lucky month or through thirty steady ones; consistency tells the two apart.

The risk columns

Leader stake — how much of the vault is the manager's own money. Skin in the game. A manager risking their own capital alongside yours is structurally more aligned than one playing with house money. But it's not a free pass — I wrote about when skin in the game still isn't enough, because a high stake on a reckless strategy just means you're both going down together.

Top depositor — how concentrated the money is. The largest single depositor as a share of the vault. If one whale is 80% of a vault, that whale's exit can pull the floor out from under everyone else. High concentration is a hidden liquidity risk wearing a normal face.

Gross leverage — real exposure, not the margin mode. Total position size divided by account value. This is the number Hyperliquid's per-asset "40×" labels don't give you: a vault can show scary per-trade margin and run modest overall leverage, or look calm and be levered to the eyeballs. Leverage is a multiplier of ruin, not just of returns — I showed exactly that in the same strategy run at 1× and 2×, і how leverage amplifies losses covers the mechanics.

The honesty flags

Some vaults carry a small amber chip. I dim those numbers rather than delete them, because why a number is untrustworthy is itself useful information:

• short-track — less than two months of history. Not enough rope to judge.
• near-zeroing — a drawdown so deep the vault nearly went to zero. The вибух drkmttr is the textbook case.
• extreme-return / extreme-calmar — numbers so large they signal a tiny base, not skill.
• solo-funded — the only depositor is the manager. No outside money has validated it yet.
• dormant — no trading volume in a month. The manager has stopped, even if the page still looks alive.

And one category the board removes entirely from the rankings: the protocol vaults (the HLP family). They're Hyperliquid's own infrastructure, not user strategies, so they're shown as a benchmark and never ranked against the vaults you can actually choose between. If you want to know what the genuinely bad actors look like, that's a different list — the vault scammer's playbook.

Why this many columns

Because no single number is safe to trust alone. A high return with a brutal drawdown is a different animal from a high return with a shallow one. A great Calmar on a dust vault is noise. A manager with huge skin in the game running 5× leverage is aligned та dangerous. The columns only mean something together — which is exactly why the board scores them for you, through three investor lenses, in the next post.

Want to see them all on live data? The board is here, every number recomputed from Hyperliquid's public API.

Third in a series. Next: Income, Balanced, Aggressive — reading the board through your own eyes.