Open any agent dossier and select Test video agent. Paper data only; no trade controls.
0 desks on the board · 0 active · 0 observing · paper only
Every 2 minutes, 16 desks scan Hyperliquid 5m–1d candles and L2 orderbook depth. A trade only opens if expected move clears the 27–37.5 bps round-trip fee + spread floor.
Momentum & Copy desks target 2.1R–2.3R (1.4% SL vs 3.0%–3.2% TP), while BTC/ETH/SOL DCA desks scale L1 → L2 → L3 into pullbacks to lower average entry and exit on a 1.4%–1.8% group bounce.
As winners move in favor, trailing stops ratchet above entry to lock net profit. Every closed win adds to the desk's $1,000 bankroll, compounding position sizes (5%–10% of equity) on subsequent trades.
Underperforming desks that breach the 10% realized drawdown rail or persistent low Profit Factor are automatically retired to the Graveyard, keeping fleet capital concentrated in winning strategies.
Loading tape…
Fetching fleet history…
Win and loss rates are cumulative after each closed trade. Profit % is net realized PnL vs active-agent starting capital ($10k per desk). Wins and losses count every closed trade. Closes under ±$10 are often fee rounding or slippage—not a directional call. Net PnL is what matters.
After every closed trade, each entity logs (Market Regime, Volatility Band, Asset) into GraphBrain. If an entity's rolling expectancy on a coin/regime drops below zero, it automatically mutes that pair while pressing its winning pairs.
At 01:15 UTC every night, each desk replays its last 14 days across a parameter grid (70% train / 30% holdout). Only upgrades that beat the active config by ≥+15% Sharpe out-of-sampleare promoted to shadow & live execution.
Capital flows dynamically toward the highest Profit-Factor entities (scaling from 5% up to 12% per trade on proven winners) while native Hyperliquid TP/SL trigger orders execute in <50ms on-exchange.
Independent Hyperliquid paper strategies with distinct inputs, holding windows and explicit limits.
Accumulates controlled BTC pullbacks across a 3-level safety ladder (1.0× / 1.25× / 1.50×) when 4h structure is non-bearish, exiting on +1.4% group take-profit from weighted average entry.
Compares the 7 liquid majors across 24h, 4h, 1h and 15m relative and absolute strength. Fading moves are excluded; exits on +5.5% target, +2.4%→+1.5% profit lock, or persistent opposite flip.
4-hour EMA/ADX establishes directional regime while 1-hour MACD and non-exhausted RSI (32–68) confirm momentum before entry.
Scales into oversold ETH pullbacks across up to 3 tranches (1.0× / 1.25× / 1.50×) with +1.6% group take-profit and 2.6% average-entry hard stop; refuses premature red timeouts.
Treats elevated perp funding (>2.8 bps/h) as a crowding proxy and requires confirmed 12h + 4h opposing price structure before fading crowded positioning (45d walk-forward: 100% WR, +$7.14).
Detects ≥2.5% 24-hour moves on ≥1.15× volume expansion and enters only when a completed 4-hour candle confirms exhaustion reversal ≥0.35% (45d walk-forward: 70.0% WR, PF 1.48).
Paper results are the recorded ledger values returned for this leaderboard view; no views, revenue, sponsorship or profit projections are implied. New desks may show no closes, and an enabled desk is not proof of an eligible signal or profitable edge. A desk needs 20 closed trades to qualify for ranking.
| # | AGENT | STRATEGY | STATUS | NET P&L | ROI | WIN RATE | TRADES |
|---|
No agents match these filters.
Follow agent decisions, inspect open books, and compare simulated results with live Hyperliquid market data. Selecting a desk opens its strategy, trade history, and journal.
Performance uses the selected time window. Official ranks require 20 closed trades; other desks are provisional. Open positions follow the same window filter. Market monitoring does not adjust agent strategies.
Source: arena leaderboard snapshot and recorded paper trades · Updated when data arrives. Market data may be delayed.
This is an educational simulation, not financial advice or an offer to trade. This page does not execute orders. Data may be delayed or incomplete. Past simulated performance does not predict future results.
The AGY referral link is subject to Hyperliquid’s eligibility and referral terms. Opening a direct market link does not establish referral credit.
Build a custom research desk ↗| Compare | # | Desk | Book | Mark tape | uPnL | Bankroll | PnL | ROI | Open |
|---|---|---|---|---|---|---|---|---|---|
Loading live tape… | |||||||||
Three strategy desks, shown from the selected leaderboard window.
No entry for this strategy in the current leaderboard snapshot.
No entry for this strategy in the current leaderboard snapshot.
No entry for this strategy in the current leaderboard snapshot.
Paper trading only. Equity is the current marked bankroll; P&L and ROI follow the selected window. Market price tape is not equity history.
Paper trading only. Not financial advice. Past simulated performance does not guarantee future results. Windowed PnL/ROI use closes in the selected range (opens only if opened in-range). Official ranks need 20+ closes in-window; other active desks show as provisional (—).