PropBlasterCockpitJournalTrade copierGuard EAReview catalogue

AI Trade Review – Review Catalog & Workflow

Part of the PropBlaster Journal. This document describes how Claude reviews your trades from screenshots, which criteria are applied, and the file interface for reviews.

How to use it

  1. When journaling, attach screenshots – ideally top-down: Daily → 4H → 1H → entry timeframe, each with entry, SL and TP drawn in. Then click Save. With the master EA's auto journal enabled, 1m / 15m / 1H screenshots are added automatically.
  2. Start the review:
  3. In the app: open the trade and click Rate trade (requires an Anthropic API key under Settings).
  4. File-based: have Claude (e.g. in a Claude project with access to your data folder) read claude/snapshot.json and the screenshots, and write the result to claude/inbox/ (format below). The app imports it within about 15 seconds. The app must be running for this; otherwise the import happens at the next start.
  5. The trade then shows the Claude review with rating, check items, detected issues and suggestions. With "Apply as mistake tags", the matching items flow into your mistake statistics. The Psychology & Mistakes tab shows the overall evaluation.

"Waiting" in the Review column means: the trade has screenshots but has not been reviewed yet. With "Request re-review" you flag a trade for another pass, e.g. after adding new screenshots.

Technical interface (file-based, in the claude/ folder of your data folder)

Format of an inbox file:

{"reviews": [{
  "trade_id": "12345678-987654321",
  "review": {
    "score": 6,
    "verdict": "One-sentence overall verdict",
    "confidence": "high | medium | low",
    "limitations": "e.g. no Daily screenshot, SL not drawn",
    "checks": [{"area": "Trend & direction", "status": "ok | warn | fail | unclear", "note": "…"}],
    "issues": ["Against higher-timeframe trend"],
    "suggestions": ["Concrete, actionable improvement"]
  }
}]}

Review catalog (general trading principles)

Every trade is checked in these eight areas. Anything that cannot be seen on the screenshots gets the status unclear. Nothing is guessed.

Area What is checked Typical fail
Trend & direction Trade direction vs. the trend on the highest available timeframe (highs and lows, higher-level structure) Short in a clear uptrend (higher highs and higher lows) or vice versa, without a visible break of structure
Market structure & timing Was the entry taken after a confirmation (break and retest, rejection at a zone, candle close) or before? Entry into a running strong counter-move ("falling knife"), entry before the break
Entry location Distance to support and resistance, position within the range Long directly below resistance, short directly above support, entry in the middle of the range or after an overextended move
Stop loss SL behind a structure point with a buffer, relative to volatility SL within normal noise (too tight), exactly on an obvious high or low, no SL
Target & R:R Is there an obstacle before the TP? Is the target realistic? R:R R:R below 1:1.5, TP behind a strong zone, no comprehensible target
Risk & position size Risk in money relative to the account balance (from the MT5 data) More than 2 % risk per trade, no SL
Management & exit Exit vs. the plan (TP/SL/manual), holding time, reason for closing Closed before the target without a reason, loss allowed to run beyond the SL
Journal consistency Do the journal entries (bias, setup, plan) match the chart? "Bias bullish" entered, chart clearly bearish

Rating (score 1–10), independent of the outcome: a cleanly executed losing trade can get an 8, a lucky rule-breaking winner a 3.

Fixed issue tags (issues), so that the evaluation stays comparable across trades: Against higher-timeframe trend · Entry without confirmation · Chased / overextended entry · Entry in middle of range · Traded into support/resistance · SL too tight · SL at obvious level · SL not based on structure · No SL · Unrealistic target · R:R too low · Closed too early · Let loss run · Over-leveraged · News risk · No clear setup · Journal contradicts chart

These tags can be applied to your mistake tags with one click: Against higher-timeframe trend → Against HTF bias, Entry without confirmation → Entered too early, Chased / overextended entry → Chased entry, SL too tight → SL too tight, Closed too early → Closed too early, Over-leveraged → Over-leveraged / too much risk, News risk → Ignored news, No clear setup → No valid setup, TP not per plan → TP not per plan.

Limitations