Bottom line up front
Only the most recent public fills are visible, so this audit covers the data covered rather than full account history. This account is loss-making in the data covered: -$122.89 realised PnL after fees on $10.3M gross volume across 135 total episodes, 17 closed. The headline obscures the real damage: three revenge trades and one averaging-down sequence consumed $260M in notional exposure and generated $260 in losses. The account is currently short $117B notional across five crypto positions (ETH short +$58.8k unrealised, SOL short +$46.7k unrealised, AVAX short +$11.4k unrealised, BTC short +$231 unrealised, ATOM long -$71k unrealised), none with stops in place. The closed-trade result is noise compared to the open leverage stack.
What the data shows
The data covered spans 1.5 days of trading (18 May to 20 May 2026) and captures 17 closed episodes. The account started with approximately $70.49M in equity and closed the window at $70.49M, a net loss of $122.89 after $64.79 in net fee drag. Gross fees paid were $139.06 against $10.3M in volume, a 1.35 basis-point take.
The closed-trade record is deeply negative. Realised PnL across all closed episodes is -$6,125.44, but the account shows only -$122.89 net loss because it is currently holding $117.3B in unrealised gains across five open positions. This is not a sign of strength; it is a sign of extreme leverage concentration. The account is short $117B notional in ETH, SOL, and AVAX, and long $71k in ATOM, all at leverage between 5x and 20x. No stops are in place on any position.
Of the 17 closed trades, the account won 8 and lost 9, a 47% win rate. The profit factor is 0.55—for every dollar won, the account lost $1.82. The average win was $18.82; the average loss was -$30.38. Expectancy per closed trade is -$7.23.
Long trades generated -$87.13 PnL on a 33% win rate. Short trades generated -$35.76 PnL on a 63% win rate. The short-side edge is marginal and fragile. Only two instruments show a "primary edge" label: INTC (short, +$134.30 on 2 episodes) and MU (long, +$9.70 on 1 episode). SILVER, SNDK, BRENTOIL, MSTR, SP500, and MRVL all show "no_edge" verdicts. The account is trading noise.
Trade quality
Win rate 47%, profit factor 0.55, expectancy -$7.23 per closed trade, win/loss ratio 0.62. These numbers mean the account loses money on average per closed trade and loses more per loss than it gains per win. The account is underwater on a per-trade basis and has no statistical edge in the closed sample.
Gross fees paid ($139.06) nearly matched the absolute value of realised losses ($6,125.44), so fees consumed 2.3% of the loss magnitude. The net fee drag of -$64.79 reflects a 61% maker rate, which is reasonable execution quality, but execution quality is irrelevant when the directional thesis is wrong.
Post-mortems
SILVER short, 18 May 23:47 to 19 May 17:58 (18.21 hours): Opened short at $78.78, closed at $75.99, -$174.13 loss on $145.6M notional (1.85x leverage). This trade is flagged as both an oversized loser (28.98x the median loss) and a revenge trade following a -$11.82 loss in MSTR. The position showed an initial adverse excursion of -0.19% and a maximum favourable excursion of +0.46%, meaning the trade briefly worked before reversing hard. The structural stop distance was 4%, so a 4% stop would have been hit at $81.45. The account did not use a stop and held into the loss.
SNDK long, 18 May 23:50 to 19 May 00:11 (0.18 hours): Opened long at $1,327.45, closed at $1,324.03, -$45.88 loss on $20.5M notional (15.4x leverage). This trade is flagged as averaging down (28 add events recorded) and an oversized loser (7.64x the median loss). The position never worked: initial adverse excursion was -0.35%, maximum favourable excursion was -0.03%. The account added to a losing position repeatedly over 10 minutes and then exited. This is textbook revenge trading after the SILVER loss.
What the risk simulation reveals
The risk simulator applies historical 1%, 2%, and 4% stop-loss rules to the closed-trade sample and projects outcomes. Under a 1% rule, simulated PnL would be +$389,671 with a deepest decline in this window of -0.62%. Under a 2% rule, simulated PnL would be +$779,343 with a deepest decline of -1.22%. Under a 4% rule, simulated PnL would be +$1,558,685 with a deepest decline of -2.38%. These are gross-of-fees figures. The simulator shows that mechanical stop discipline would have turned this loss-making sample into a multi-million-dollar profit. The account's actual behaviour—holding losers and adding to them—is the inverse of what the data rewards.
Open positions
The account holds five open positions totalling $117.3B notional in unrealised gains, but this figure is misleading because it reflects extreme leverage and no downside protection.
ETH short dominates: $58.9B notional unrealised gain at 15x leverage, entry $2,121.18. No stop in place.
SOL short: $46.7B notional unrealised gain at 20x leverage, entry $84.67. No stop in place.
AVAX short: $11.4B notional unrealised gain at 10x leverage, entry $9.61. No stop in place.
BTC short: $231 unrealised gain at 20x leverage, entry $76,824.90. No stop in place.
ATOM long: -$71k unrealised loss at 5x leverage, entry $2.04. No stop in place.
The unrealised PnL across these positions is still open and not a locked realised result. The account is short $117B notional in three major cryptocurrencies with no stops. A 1% move against these positions would erase the entire unrealised gain and trigger liquidation risk.
Honest summary
- Strength: Short-side trades show a 63% win rate versus 33% on longs, suggesting the account has identified a directional bias that works in the available sample. The INTC short (+$134.30) and the BRENTOIL short (+$3.35) both closed profitably.
- Weakness: Revenge trading and averaging down are endemic. Three consecutive revenge trades (MSTR, SILVER, BRENTOIL) opened within 8 minutes of each other and consumed $260M in notional. The SNDK long was added to 28 times over 10 minutes while underwater. The account