Trades marked as bugs are excluded from all summary statistics (WR, PnL, Sharpe). They remain visible in the journal at reduced opacity for transparency. Each bug was identified, root-caused, and fixed.
0. First-day initialization (1 trade, May 22)
SOL short opened before exit parameters were configured. Held 0 bars (5 seconds), no SL/TP set, immediate timeout. Lost $24 in fees on a zero-move round-trip.
1. Pre-Strategy C trades (9 trades, May 23)
QUAKE cascade trades before the quality scoring system (Strategy C) was implemented. These fired on low-quality B/C-tier cascades that would be filtered by the current scorer. Fixed: Strategy C quality gate now requires score >= 130 (A-tier+).
2. First QUAKE trades (5 trades, May 28-31)
Early QUAKE positions opened before exit parameters (trailing 9%/2.5%, timeout 44h) were optimized. Used default exits that were too loose. Fixed: walk-forward validated exit params deployed Jun 2.
3. 3x cascade execution bug (7 trades, Jun 5-7)
Multi-cascade guard compared against oldest position timestamp instead of most recent. Same S-tier cascade executed 3 times (18 positions instead of 6). HL merged multiple shorts per coin — can't fully separate legit from bug trades. Fixed: guard now compares most recent cascadeTs, max 12 total positions.