trade-journal-analyzer

name	trade-journal-analyzer
description	Unified post-trade analytics: journal pattern extraction + drawdown classification. Absorbs: drawdown-classifier.
argument-hint	analyze journal | patterns | edge audit | classify drawdown | is this noise or real
allowed-tools	["Read","Bash"]
auto-invoke	true
model	default
context_trigger	analyze trades, journal patterns, win rate, edge audit, losing streak, drawdown, is my system broken, trade review

Trade Journal Analyzer (Expanded)

Absorbs: drawdown-classifier

Reads historical trade entries, extracts actionable statistical patterns, AND classifies drawdowns. Converts a "diary" into a "data warehouse."

Triggers

"analyze my trades", "journal patterns", "what's my actual WR", "edge audit", "losing streak", "drawdown", "is my system broken", "should I stop trading", "3 losses in a row"

Core Analytics

1. Ingest: Read entries from .context/trading_journal/.
2. Parse: Extract setup type, instrument, direction, entry, SL, TP, result, notes.
3. Analyze:
   • Win Rate by Setup Type: Which setups are actually profitable?
   • Win Rate by Instrument: Where is the edge strongest?
   • Win Rate by Time of Day: Asian vs London vs NY session performance.
   • Average R:R Achieved: Planned RR vs actual RR (execution gap).
   • Drawdown Sequences: Longest losing streaks, recovery time.
   • Edge Decay: Is WR trending up or down over last 20 trades?
4. Flag:
   • Setups with WR < breakeven threshold → flag for review or removal.
   • Instruments with consistent negative EV → stop trading them.
   • Emotional notes correlation → do emotional trades have lower WR?

Drawdown Classification

Not all drawdowns are equal. The wrong response is more dangerous than the drawdown itself.

Class 1: Noise (Random Variance)

• Losing streak within expected statistical bounds for the system's WR.
• Test: At 60% WR, a 5-loss streak has P = 0.4^5 = 1.02%. Over 200 trades, ~2 expected.
• Response: Do nothing. Continue executing. Do NOT adjust.

Class 2: Structural (Setup Flaw)

• Losses concentrated in a specific setup, instrument, or time period.
• Test: Is the WR decline isolated to one setup type?
• Response: Quarantine the specific setup. Continue trading others.

Class 3: Thesis-Breaker (Edge Invalidation)

• Systematic WR decline across ALL setups.
• Test: Is the WR decline persistent (>30 trades)? Has market microstructure changed?
• Response: Full stop. Trigger circuit-breaker. Paper trade. Re-validate edge.

Output Format

## Trade Journal Analysis (Last N Trades)

### Win Rate by Setup
| Setup | Trades | Wins | WR | Avg RR | EV/Trade |
|-------|--------|------|----|--------|----------|

### Win Rate Trend (Rolling 20)
[Trending UP / DOWN / FLAT] — current WR: XX%

### Drawdown Classification
Observed: X losses in last Y trades
P(this streak | WR=Z%): XX.X%
Classification: [NOISE / STRUCTURAL / THESIS-BREAKER]
Prescribed Action: [Continue / Quarantine setup X / Full stop]

### Edge Health
Verdict: [HEALTHY / DECAYING / CRITICAL]

Integration

• Feeds into zenith-execution for forward simulation (Monte Carlo)
• Validates Kelly assumptions (is the stated WR real?)
• Triggers circuit-breaker if edge decay is CRITICAL