Trading Journal — 交易日志管理

Record and analyze futures/stock trading journals. Supports structured trade logging, P&L tracking, performance analytics, and multi-format export.

Architecture

Trade Entry → [Journal] → Log → [Analysis] → Stats / Reports
                 ↑                        ↓
           Template Driven          Export (MD/CSV)

Core Workflows

Workflow 1: Log a Trade Entry

Use the standardized template to record a trade:

Template fields:

Full JSON template:

{
  "date": "2026-05-20",
  "symbol": "IF2606",
  "direction": "long",
  "entry_price": 3850.0,
  "exit_price": 3900.0,
  "quantity": 2,
  "multiplier": 300,
  "entry_time": "09:35",
  "exit_time": "14:20",
  "fees": 45.60,
  "strategy": "trend_follow",
  "notes": "突破前高入场，尾盘止盈",
  "tags": ["股指", "日内", "趋势"]
}

To log a trade:

1. Ask user for trade details, or accept them in free-text format
2. Map to the standard template
3. Append to the daily journal file via the scripts/trade_logger.py script
4. Calculate P&L and confirm with user

Workflow 2: Calculate P&L

Auto-calculated based on template fields:

Futures P&L:

P&L = (exit_price - entry_price) × quantity × multiplier × direction_sign - fees

Stock P&L:

P&L = (exit_price - entry_price) × quantity - fees

Where direction_sign = 1 for long, -1 for short.

The system handles:

• Open trades (no exit price — show unrealized P&L using current market price)
• Partial fills (quantity adjustments)
• Multi-leg strategies (aggregate P&L)

Workflow 3: Performance Analysis

Run analytics on journal entries using scripts/trade_analyzer.py:

Metrics computed:

Analysis frequency:

• Daily: Quick summary of today's trades
• Weekly: Performance trends, strategy comparison
• Monthly: Full analytics report
• Custom: Any date range specified by user

Workflow 4: Archive & Search

Journal files are organized by date:

journals/
├── 2026/
│   ├── 2026-05/
│   │   ├── 2026-05-18.json    # Day trades
│   │   ├── 2026-05-19.json
│   │   └── 2026-05-20.json
│   └── 2026-06/
│       └── 2026-06-01.json
└── archive/
    └── 2025-journal.jsonl

Search across journals by:

• Date range (YYYY-MM-DD to YYYY-MM-DD)
• Symbol (e.g., IF, IC, CU)
• Direction (long/short)
• Strategy tag
• P&L range
• Custom text search in notes

Workflow 5: Export

Export journals to preferred format:

Usage Guide

A) Recording a New Trade

When user says "记一笔交易" or provides trade details:

1. Parse trade info from user's natural language
2. Confirm parsed fields with user
3. Run: python scripts/trade_logger.py add \x3Cjson_string>
4. Show calculated P&L and confirmation

Example interaction:

User: 今天IF2606做多2手，3850进场，3900平仓
Agent: 📓 记录交易：
  IF2606 | LONG | 3850 → 3900 | 2手
  P&L: (3900-3850)×2×300 - 费用 = ¥29,954.40
  确认记录？(是/否)

B) Checking Performance

When user asks "看看收益" / "我的交易怎么样":

1. Determine time range (today/this week/this month/custom)
2. Run analyzer: python scripts/trade_analyzer.py --range 2026-05-01 2026-05-20
3. Present formatted results

Example output:

📊 交易绩效报告 (2026-05-01 ~ 2026-05-20)
━━━━━━━━━━━━━━━━━━━━━━━━━━━
  总交易数:  24
  胜率:      70.83% (17/24) 🟢
  总盈亏:    +¥186,520.00 🟢
  平均盈利:  +¥16,250.00
  平均亏损:  -¥4,830.00
  盈亏比:    3.36
  最大回撤:  -¥12,400.00
  最佳交易:  +¥45,000.00 (IF2606, 05-08)
  最差交易:  -¥8,200.00 (IC2609, 05-12)

C) Exporting Data

When user asks to export:

1. Confirm format and date range
2. Run export script with appropriate parameters
3. Show export path and summary

Scripts

scripts/trade_logger.py

Core trade logging — add, edit, delete, and list journal entries.

scripts/trade_analyzer.py

Performance analysis — compute metrics, generate reports for any date range.

scripts/journal_export.py

Multi-format export — output journals to Markdown, CSV, or JSON.

Special Notes

• Journal files are stored in journals/ directory within this skill folder
• Each day gets one JSON file; multiple trades append to the same file
• P&L is calculated on entry, not just on display
• Open positions are tracked with exit_price: null and open: true
• All monetary values are in CNY (¥)
• Fees should include commission + slippage + stamp duty where applicable
• Journal data is local-only — no external API calls