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 -46.43% in the data covered, a loss of $115,165 on a starting balance of roughly $248,057. The headline pattern is unambiguous: one catastrophic BTC long position ($3.95M notional, -$87,638 loss) opened as a revenge trade after an ETH loss, followed by serial oversized positions in illiquid alts and repeated averaging down on ETH. The highest balance in this window was $274,804 on 15 May; the lowest balance in this window was $153,255 on 18 May, a deepest decline in this window of -44.23%. Fees consumed $30,962, but the real damage was position sizing and the absence of any structural risk management.
What the data shows
The account opened on 15 May and closed 34 trades in four days. The arc is sharp: early wins on SOL and HYPE were immediately overwhelmed by the BTC catastrophe. On 15 May at 05:31, after a $1,030 loss on ETH, the account opened a BTC long at $3.95M notional with 25x leverage. This position ran for 66 hours and exited on 17 May at $78,717.57, crystallising -$87,638. That single trade consumed 76% of total realised losses.
Immediately after the BTC exit, the account opened a revenge trade in ETH at $842,667 notional. This pattern repeats: after the BTC loss, ETH long at $886,943 notional; after an ETH loss of -$572, another ETH long at $730,402 notional. The behavioural flags record five revenge trades across the window, all in ETH or BTC, all following losses.
The coin-by-coin breakdown shows no edge anywhere. BTC: 0% win rate, one episode, -$87,638. ETH: 0% win rate across 15 episodes, -$15,011 total. HYPE: 33% win rate, -$4,288 net. ONDO: 0% win rate, -$3,207. Only TON produced a win: +$2,257 on a single 100% win-rate trade. The long-only bias is total: $0 in short PnL, all 34 closed trades were longs.
Fees paid were $30,962 on $123.18M gross volume, a 2.5bps all-in rate. But fees are noise relative to the position-sizing failures. The account averaged -$3,387 per closed trade (expectancy). The profit factor is 0.04: for every dollar won, the account lost $25.
Trade quality
Win rate is 14.71%, meaning roughly 1 in 7 trades closed profitably. The profit factor of 0.04 and win/loss ratio of 0.21 are terminal metrics: average wins of $856 cannot offset average losses of $4,119. Expectancy of -$3,387 per trade is the core finding—this account has negative expected value on every entry.
The longest winning streak was one trade. The longest losing streak was 13 consecutive closed trades. No structural stops were in place on the two largest losses; the BTC trade has no entry price recorded, suggesting a market order or rapid scaling into a position. The ONDO trade had a 4% structural stop available but exited at entry price, indicating the stop was either not used or the position was closed at breakeven despite a -2.91% maximum adverse excursion.
Post-mortems
BTC long, 15 May 05:31 to 17 May 05:46 (66 hours): Opened at $3.95M notional with 25x leverage, exited at $78,717.57, loss of -$87,638. Flagged as both oversized loser (85x the median loss) and revenge trade (opened after -$1,030 ETH loss). No entry price recorded. This is the defining trade of the window. The position was held for nearly three days and represents a single catastrophic bet that consumed the account's edge for the entire period.
ONDO long, 16 May 02:27 to 18 May 12:00 (45.5 hours): Opened at $0.35, exited at $0.35, loss of -$3,207 on $228,003 notional. Flagged as oversized loser (3.1x median). Despite entry and exit at the same price, the position experienced a -2.91% maximum adverse excursion and a +1.95% maximum favourable excursion, indicating intra-trade volatility. A 4% structural stop was available but not triggered. This is a second-order loss, but the position size relative to account equity was reckless.
What the risk simulation reveals
Under a 1% hard stop rule applied historically, the account would have realised -$4,071 with a deepest decline in this window of -2.85%. Under 2%, the loss would have been -$8,143 with a deepest decline in this window of -5.7%. Under 4%, the loss would have been -$16,285 with a deepest decline in this window of -11.4%. All three simulations produced 0% win rate, meaning stops would have prevented catastrophic losses but would not have created edge—they would have simply capped the damage. The actual deepest decline in this window of -44.23% shows the cost of unmanaged risk.
Open positions
Five open positions remain, all longs, all without stops in place. ETH long at 25x leverage with -$306.71 unrealised loss is the largest open exposure. AAVE long at 10x with -$56.80 unrealised loss. UNI, PENDLE, and NEAR are small and slightly positive. None have stop orders. The ETH position carries the same leverage and coin as the revenge trades that followed the BTC loss, suggesting the pattern may still be active.
Honest summary
- Visible strength: The account can execute at scale and move quickly. The 1.64% maker rate and $123M gross volume in four days show operational competence.
- Visible weakness: Position sizing is entirely decoupled from account equity or risk tolerance. The BTC trade at $3.95M notional on a $248K starting balance is a 16x account bet. No trade had a structural stop in place. Revenge trading is explicit in the data: five separate instances of opening large positions immediately after losses.
- Visible weakness: Long-only bias with no short-side edge. 34 closed trades, all longs, 14.71% win rate, negative expectancy on every trade. The account is not unlucky; it has no edge.
- Data scope caveat: Only the most recent 10,000 fills are visible. This window spans four days in May 2026. Earlier account history, if it exists, is not covered by this audit.