Bottom line up front
Only the most recent public fills are visible, so this audit covers the data covered rather than full account history. The account is profitable in that window: $313,586 realised PnL across 23 closed episodes, but the headline masks severe structural fragility. A single oversized long position on ETH opened 3 November cost $1.18m—nearly four times the median loss—and consumed all gains from the short-side edge. The short book is genuinely profitable (90% win rate, $788k realised), but long-side attempts and revenge-driven averaging have been destructive. The highest balance in this window reached $8.45m on 24 December; the lowest balance was $48.5k on 22 October. The deepest decline in this window was 81.35%, driven almost entirely by one position that was allowed to run against a structural stop only 1.14% away.
What the data shows
This is a short-biased account that found a genuine edge in ETH shorts but has repeatedly undermined it through long-side re-entries and position sizing discipline failures. The data covered spans 73 days from 14 October 2025 to 27 December 2025.
The account opened with a strong short: ETH short from 4127.01 to 3984.43 (14–30 October), closed for $881k profit with 77 averaging-down events and a maximum notional of $16.5m. This trade established the pattern: the short side works, sizing is aggressive, and averaging is the default response to adverse movement. The account then cycled through three profitable re-entries into ETH shorts on 31 October, 31 October again, and 2 November, each closed within hours to days for $13k, $7k, and $16k respectively. These were FOMO re-entries following previous closes, and they worked.
The structural break came on 3 November. A long position opened at 3820.22, accumulated to a maximum notional of $15.47m across 54 averaging events, and held for 569 hours before closing at 3137.8 on 26 November for a loss of $1.18m. This single trade consumed the entire profit buffer from the short book and then some. The position had a structural ATR-based stop only 1.14% away—tighter than any other position in the window—yet was allowed to run through it. The averaging pattern is visible: the account added at 3836.4 repeatedly, fighting the move downward.
After the loss, the account recovered sharply. A long from 2879.67 to 3007.4 (27 November–2 December) generated $589k profit, the second-largest win in the window. This was followed immediately by a short opened 2 December at 3062.85, which also averaged down (54 events, $19.3m notional) and closed 26 December at 3168.81 for a loss of $200k. The short-side loss is smaller in absolute terms but reveals the same pattern: averaging into losers, ignoring structural stops (1.56% away), and holding through the pain.
Realised PnL is $351k gross, but net fee drag of $37.8k (10.83% of realised PnL) is material. The account paid $38k in explicit fees across $280m in gross volume, with 68% maker fills. The fee ratio is not catastrophic, but it is a meaningful tax on an edge that is already thin once the long-side losses are netted.
Long versus short: shorts generated $788k across 10 episodes (90% win rate), while longs generated a loss of $474k across 13 episodes (84.62% win rate). The win rate on longs is deceptively high because most long episodes are small or quick; the two large long positions (3 November and 2 December) account for the entire loss. The short side is the genuine edge; the long side is where discipline breaks.
Trade quality
Win rate of 86.96% across 23 closed episodes is strong on the surface. Profit factor of 1.23 means every dollar of losses is offset by $1.23 of wins—acceptable but not exceptional. Expectancy of $13,634 per closed episode is positive but modest given the notional sizes deployed. The win/loss ratio of 0.18 is the red flag: the average win is $84.8k, but the average loss is $460.8k. This is a distribution problem. The account wins frequently but loses catastrophically when it does lose. The maximum loss streak is 1, which is fortunate; a second consecutive large loss would have been terminal.
Post-mortems
ETH long, 3 November 2025 to 26 November 2025. Opened at 3820.22, closed at 3137.8, loss of $1.18m over 569 hours. Maximum notional $15.47m. This position was flagged for averaging down and oversized loss. The structural ATR-based stop was set at 1.14% below entry, yet the position was held through a 17.8% decline. The averaging pattern shows 54 separate add events, predominantly at 3836.4, fighting the downtrend. This is the defining failure of the data covered. The position size relative to the loss magnitude (5.88x the median loss) indicates either a sizing error or a complete breakdown in stop discipline. There is no evidence the stop was ever triggered or respected.
ETH short, 2 December 2025 to 26 December 2025. Opened at 3062.85, closed at 3168.81, loss of $200.9k over 563 hours. Maximum notional $19.29m. This position was flagged for averaging down. The structural ATR-based stop was 1.56% away; the position moved 3.46% against the entry before being closed. The averaging pattern shows 54 add events, suggesting the account was fighting a rally in the same manner it had fought the decline in the November long. The loss is smaller than the November disaster, but the behaviour is identical: averaging into a losing position, ignoring the structural stop, and holding until forced to capitulate.
What the risk simulator reveals
Under a 1% hard stop rule applied historically to all episodes in the data covered, the account would have realised $79.5k profit with a deepest decline of 53.57%, stopping out of 4 episodes early. Under a 2% rule, the simulated result would be $159k profit with a deepest decline of 71.88%, also stopping 4 episodes early. Under a 4% rule, the simulated result would be $318k profit with a deepest decline of 86.7%, stopping 4 episodes early. The 4% simulation nearly matches the actual result, which is unsurprising given that the two largest losses (November long and December short) both ran well beyond 4% before closing. A 1% rule would have prevented the November catastrophe entirely and would have reduced the deepest decline from 81.35% to 53.57%—a material improvement in capital preservation.
Open positions
No open positions at the time of the latest fill on 27 December 2025.
Honest summary
- Short-side edge is real. The 90% win rate and $788k profit on shorts across 10 episodes is not noise. The structural stops on winning shorts are tighter (3.0% on the October win, 1.54% on the November short win) and are being respected. The short book works.
- Long-side discipline is absent. Both large long positions (November and December) were allowed to run through structural stops and were averaged into aggressively. The account has a 84.62% win rate on longs, but this is because most long episodes