Dissecting Hyperliquid's Top Profitable Addresses: What Winners Actually Pay in Fees

Recent analyses dissected tens of thousands of Hyperliquid addresses to answer how winners trade. We flip to a lens almost nobody uses: how much do the winners pay to trade?

This is possible because Hyperliquid's fills are fully public and every fill carries the actual settled fee — no estimation needed. All numbers below come from the public API on 2026-09-02; reproduction steps are at the end.

The big picture — with a health warning

Of 44,348 leaderboard addresses, 22,248 (50.2%) show positive all-time PnL, and 59.6% are positive over the last 30 days.

Far better than the folk wisdom that 90% lose — but this is a survivorship sample: the leaderboard only contains accounts still alive and trading. It says active accounts often win; it does not say half of everyone wins.

Method (transparent, reproducible)

1. Load all 44,348 leaderboard rows (account value; day/week/month/all-time PnL, ROI, volume);
2. Filter: all-time PnL>0 and ROI>0, account value ≥ $100k, 30-day volume ≥ $1M — 2,770 qualify;
3. Take the top 15 by 30-day PnL and pull each address's latest 2,000 public fills (30,000 fills total), summing the real fee field.

Finding 1: winners pay a third of the standard rate

The 15 addresses did $238.9M of sample volume for $42,236 in net fees — 1.77 bps average, ~1.4 bps median, range 0.53–4.7 bps.

Benchmark: the standard taker tier is 4.5 bps, which would cost ~$107,500 on the same volume. The winners paid 60% less.

Finding 2: 9 of 15 are maker-dominant

By the taker flag in fills, 9/15 addresses cross the spread on fewer than half their fills; three are near-pure makers (0%, 5%, 8% taker).

High-turnover edges are tiny (public analyses put them around 0.2–0.45% of volume) — fees live in the same order of magnitude. Fee management is not a savings tip; it decides whether the strategy exists at all.

Finding 3: the largest account gets paid on two-thirds of its fills

The top address by account value (~$59.3M) did its latest 2,000 fills in 0.85 days, trading $51.9M:

• fees paid: $2,389;
• maker rebates received: $1,163 (62% of fills had negative fees);
• net: $1,226 — 0.24 bps.

The same volume at the standard taker tier is $23,354. Top players trade the same market at ~20x lower cost than a default account.

What a regular trader can copy

Bluntly: not the volume tiers. But three things transfer:

1. Audit your own real rate — userFills works for your address too; sum your fees over volume and learn what you actually pay;
2. Post, don't cross, when your strategy allows — the maker/taker gap is structural and compounds;
3. Use every discount you can reach — staking discounts and referral channels have no volume gate. Registering via a rebate route gets you a fee discount: entry here.

Honest limits

• 2,000-fill windows span 0.02–280 days across addresses — not a like-for-like time comparison;
• Cross-venue hedges are invisible; profitability is leaderboard-defined;
• Fee tiers/staking are inferred from outcomes, not known;
• This is a cost-structure piece, not trading advice — copying positions without the system behind them is meaningless.

Data & reproduction

• Leaderboard: stats-data.hyperliquid.xyz/Mainnet/leaderboard (pulled 2026-09-02, 44,348 rows)
• Fills: api.hyperliquid.xyz/info userFills (latest 2,000 per address, real fee field)
• Our raw per-address results are archived and open to review

Rates are actual-paid estimates from public data; your account may differ. Not investment advice.