Trader types
Winners by trader type
Each wallet gets the first rule it matches, top to bottom. The last column shows how common the same type is among wallets that lost more than $100K.
| Type | Rule | Wallets | Share of PnL | Median PnL | Median hold | Orders / day | Median drawdown | Human | Hybrid | Algo | Share of losers |
|---|
Humans and algos
The remaining — of winners placed too few orders since Jul 2025 to classify.
Share of wallets and of profit
Algos are a minority of winners but earn far more each. Among big losers they are rarer still.
When they trade (hour of day, UTC)
Average wallet's share of its orders placed in each hour. Algos run around the clock; humans peak with the US session.
The bot score blends eight signals: orders seconds apart, hours active per day, orders per day, peak orders per minute, round-the-clock spread, client order ids (set by API code), sizes that are not round numbers, and no frontend builder code. Algo = score ≥ 0.6, human < 0.35. A slow algo looks human here, and a human using the UI's scale-order tool can look hybrid.
How fast they trade
Holding time against trading rate, one dot per winner
Median round trip (flat to flat) on the vertical axis, orders per active day across. Both log scales.
What they trade
Volume mix by trader type
Average wallet's share of volume. Majors = BTC, ETH, SOL, HYPE perps. TradFi = HIP-3 stocks, indices, commodities, FX.
Drawdown risk
Worst drawdown as a share of peak profit
Bar spans the middle half of wallets (25th to 75th percentile); the dot is the median. Measured on the daily realized-PnL-plus-funding curve, so open losing positions are not counted.
Winners against losers
Median of each trait. The clearest gaps are resting orders, liquidations and win rate.
| Trait | Winners | Losers |
|---|
Find the wallets that win
AlphaHunter screens millions of on-chain traders by realized PnL, win rate and holding time, so you can study and follow the few who are consistently ahead.
Get StartedWho's counted. Every Hyperliquid wallet that made more than $100K between Mar 22, 2025 and Oct 18, 2026, with at least 30 active days across 3 or more months. Vaults (including HLP) are excluded. The comparison group is the wallets that lost more than $100K under the same rules.
Profit. Perp profit after fees, plus funding, plus open perp positions at the last trade price. Spot profit counts only when the wallet's cost basis is known.
Human or algo. A score built from trading behaviour since Jul 2025: orders placed less than a second apart, hours active per day, orders per day and per minute, round-the-clock activity, API order ids, non-round sizes and frontend builder codes. It is an estimate. A slow bot can look human, and a human using scale orders can look hybrid.
Drawdown is measured on daily realized profit and funding, so losses on positions that are still open are not counted. One person can run several wallets, and each counts separately. Nothing here is financial advice.
Source: Hyperliquid node fill, funding and ledger data, analysed by AlphaHunter.