- Open ETH short dominates this wallet: $34,095,297 notional, -$607,348 -27.2% unrealised, 15x cross, held at least 52 days, liquidation $3,255.
- Closed-trade context: -$110 realised trading PnL across 13 closed position cycles in the data covered.
- Data used: latest 10,000 public fills from May 19, 2026 to Jul 10, 2026; older public fills may exist outside this audit because the source hit its cap.
- The hold duration is a lower bound because the position was already open at the first visible fill for that market.
@Wintermute - 0xecb63caa47c7c4e77f60f1ce858cf28dc2b82b00
@Wintermute wallet audit
Open ETH short dominates this wallet: $34,095,297 notional, -$607,348 -27.2% unrealised, 15x cross, held at least 52 days, liquidation $3,255. Closed trades are supporting context: -$110 realised trading PnL across 13 closed position cycles, using the latest 10,000 public fills from May 19, 2026 to Jul 10, 2026; older public fills may exist outside this audit.
Closed trades still matter, but they are not the main account story here. The closed-trade sample covers May 19, 2026 to Jul 10, 2026; the open-position figures are live account-state figures from Hyperliquid when the audit ran. The unrealised PnL is still open, not a locked result; liquidation at $3,255 remains the downside boundary for this position.
- Position
- ETH short
- Open PnL
- -$607,348 (-27.2% ROE)
- Notional
- $34,095,297
- Liquidation
- $3,255
- Read this as
- entry $1,762 · mark $1,794 · held at least 52 days · 15x cross · $2,273,020 margin used
This audit is position-dominated, so open unrealised PnL is shown separately from closed realised trading PnL. The dollar PnL is the realised result from closed trades in the data covered. The percentage uses an inferred starting value (current account value $90,996,837 minus closed trading PnL -$110 = starting estimate $90,996,947). This audit does not ingest a deposit or withdrawal ledger, so it can show that trades lost money, but it cannot prove whether the owner also moved funds in or out. Older fills may also exist outside the latest 10,000-fill window.
This is not a fixed last-week or last-month period. It is the actual span covered by the latest 10,000 public fills Hyperliquid exposed for this wallet. Because the public fill source hit its cap, older trades may exist but are not included here.
- Public fills
- 10,000
- Position cycles
- 13 closed, 130 open
- Limit
- latest 10,000 fills only
- SOL short is genuinely profitable: +$226,594 unrealised on a 52-day hold. The trade is working, the sizing is appropriate for a winner, and the liquidation price is far enough away to absorb volatility.
- Short bias has a 60% win rate on closed trades. The three closed shorts won twice. Longs are 12.5% win rate. The directional bias is sound; the execution on individual longs is not.
- You're still solvent and the account is at $91m. The highest balance in this window was $161.1m on June 17th; the lowest was $68.7m on May 20th. The deepest decline in this window was -48.72%. You survived the draw.
The opening paragraph
You're short 19,007 ETH at 1761.84, holding a $34.1m position on 15x leverage since May 19th, down $607,348 unrealised—that's a -27.2% ROE on a single trade that's still bleeding. The liquidation boundary sits at $3,255, which means you need ETH to nearly double from here before the position gets force-closed, but the unrealised loss is very much real and unlocked. Only the most recent public fills are visible in this data covered, so this snapshot captures the open book as of the analysis date, not a final outcome.
The greatest hits
- SILVER revenge trade on May 19th: Lost $161.99 on a $145,646 position in 12 hours after taking a $35 loss on SP500 minutes earlier. The position was 37.8x the median loss size. This is textbook revenge sizing.
- SP500 and BRENTOIL stacked on the same day: Two oversized longs ($67,236 and $27,795 notional) both opened and closed on May 19th for -$35 and -$28 respectively. Combined they cost $64 in a 12-hour window while the account was already underwater.
- The ETH short itself: Entered at $1,761.84 on May 19th, now at $1,793.80 mark price. The position has moved against you for 52 days straight. Funding data-covered is -$7.38m gross; funding since open is -$110,281. You're paying to hold this trade.
- Closed realised PnL is -$3,888 after $8,419 in net fee drag. The 13 closed episodes generated $109.70 in raw profit before fees, then fees erased it and then some. Profit factor is 0.56—you're losing $1.44 for every $1 you make on closed trades.
- SOL short is the only bright spot in the open book: +$226,594 unrealised on a $14.9m notional short at 20x leverage, held 52 days. It's a +29.93% ROE position, but it's drowning in a sea of red. The liquidation boundary is $222.79.
The pattern
The closed trades on May 19th show a trader chasing losses in real time: SP500 loss → SILVER revenge entry at 5x the median position size → BRENTOIL and MSTR piled on the same day. The open book is dominated by two mega-shorts (ETH and SOL) that have been held for 52 days with cumulative funding costs in the millions. The account is net short crypto and commodities, but the sizing is inverted—the losing position (ETH) is the largest, the winning position (SOL) is secondary. Fee drag is eating the closed-trade edge alive.
The reluctant compliments
- SOL short is genuinely profitable: +$226,594 unrealised on a 52-day hold. The trade is working, the sizing is appropriate for a winner, and the liquidation price is far enough away to absorb volatility.
- Short bias has a 60% win rate on closed trades. The three closed shorts won twice. Longs are 12.5% win rate. The directional bias is sound; the execution on individual longs is not.
- You're still solvent and the account is at $91m. The highest balance in this window was $161.1m on June 17th; the lowest was $68.7m on May 20th. The deepest decline in this window was -48.72%. You survived the draw.
The verdict
You're holding a $34m short that's cost you $607k in unrealised losses and $7.4m in data-covered funding, while your closed trades lost money after fees and your revenge entries on May 19th read like a panic spiral. The SOL short is the only thing keeping the account from looking catastrophic, and it's half the size of the ETH anchor around your neck.
Behaviour checksRule-based warnings found in the trading history. They are not moral judgements; they mark patterns worth reviewing.
Rule-based position-cycle checksNo matching position cycles in the data covered.
No matching position cycles in the data covered.
- xyz:SP500: -$35 realised loss; 8.3x median closed loss.
- xyz:SILVER: -$162 realised loss; 37.8x median closed loss.
- xyz:SILVER on May 19, 2026: followed a -$35 loss; larger-than-normal size.
- xyz:INTC on May 19, 2026: followed a -$162 loss; larger-than-normal size.
Expectancy is not a forecast. It is the historical average result per closed position cycle in this reconstructed sample.
Risk simulatorA counterfactual replay of the same historical trades using fixed risk limits. It is for comparing risk shape, not predicting future returns.
Replays the same closed position cycles with 1%, 2%, and 4% account-risk sizing. It shows what the wallet would have made or lost if each eligible cycle was sized from account value at entry and a structural stop.
- Max drawdownLargest high-to-low account-value drop inside this simulated replay.
- -1.2%
- Stopped earlyHow many historical position cycles would have exited before the real close because the simulated stop was hit.
- 2
- Max drawdownLargest high-to-low account-value drop inside this simulated replay.
- -2.4%
- Stopped earlyHow many historical position cycles would have exited before the real close because the simulated stop was hit.
- 2
- Max drawdownLargest high-to-low account-value drop inside this simulated replay.
- -4.7%
- Stopped earlyHow many historical position cycles would have exited before the real close because the simulated stop was hit.
- 2
The 1%, 2%, and 4% rules are account-risk limits per position cycle, not leverage settings. If the simulated stop is breached, the cycle is stopped early. Outputs are gross of fees and funding, so use them as risk-shape comparisons rather than exact alternate realised trading PnL.