Automated China A-share & HK stock analysis system powered by OpenClaw gateway. Combines three investment methodologies (CANSLIM, Value Investing, Growth) with macro cycle analysis, delivering actionable insights via WeChat.
- Multi-Master Scoring — CANSLIM (O'Neil) + Value (Duan Yongping) + Growth (Peter Lynch)
- Market Screener — 全市场 CANSLIM 扫描 (中证800 + 创业板综), 双路线筛选 (底部蓄势 + 强势创新高), 6 个基本面反转信号 + 走势确认 + 爆量豁免
- Technical Analysis — MA, RSI, MACD, Bollinger Bands, volume analysis
- Fundamental Analysis — PE, PB, ROE, PEG, 10-year financial history
- Macro Cycle Analysis — Howard Marks 14-dimension market cycle positioning
- Trading Signals — Multi-dimension buy/sell signals with pattern recognition
- Automated Reports — Morning briefing, midday update, closing summary, weekly report
- Price Alerts — Configurable thresholds checked every 30 minutes during trading hours
- WeChat Delivery — Auto-split long messages, supports WeCom App/Bot, Server酱, PushPlus, Telegram
- Docker & Docker Compose
- An OpenClaw account and gateway token
git clone <repo-url> && cd openclaw-docker
# Copy example config
cp openclaw.example.json data/openclaw.json
# Edit data/openclaw.json — set your gateway token
# Configure notifications
cp data/stock-data/notifier.example.json data/stock-data/notifier.json
# Edit data/stock-data/notifier.json — set your WeChat webhook key
# Enable security hook
git config core.hooksPath .githooksdocker compose up -dDependencies are installed into the host-mounted ./pip-user directory so they
survive container rebuilds (the directory is mapped to /home/node/.local):
# pip is bootstrapped from Debian repos (use root)
docker exec --user root openclaw apt-get update && \
docker exec --user root openclaw apt-get install -y --no-install-recommends python3-pip
# Install project deps as the `node` user (writes to ./pip-user/)
docker exec --user node openclaw \
python3 -m pip install -r /home/node/.openclaw/skills/china-stock/requirements.txtEdit data/stock-data/watchlist.json to add your stocks:
{
"groups": {
"核心持仓": ["600519", "000858"],
"科技成长": ["300750", "hk00700"],
"金融地产": ["601318", "000002"]
}
}# Run a closing report
docker exec -w /home/node/.openclaw/skills/china-stock openclaw python3 runner.py close
# Run market cycle analysis
docker exec -w /home/node/.openclaw/skills/china-stock openclaw python3 market_cycle.py
# Force refresh + send to WeChat
docker exec -w /home/node/.openclaw/skills/china-stock openclaw python3 market_cycle.py --refresh --send
# Run full-market CANSLIM screener (first run ~5min, cached run ~40s)
docker exec --user node -w /home/node/.openclaw/skills/china-stock openclaw \
python3 runner.py screen
# Dry-run with 3 sample stocks (smoke test)
docker exec --user node -w /home/node/.openclaw/skills/china-stock openclaw \
python3 runner.py screen --dry-run| Job | Time | Command |
|---|---|---|
| 早盘简报 | 9:00 Mon-Fri | python3 runner.py morning |
| 午盘快报 | 11:35 Mon-Fri | python3 runner.py midday |
| 收盘总结 | 15:05 Mon-Fri | python3 runner.py close |
| 价格预警 | Every 30min (trading hours) | python3 runner.py alert |
| 周报 | Sat 10:00 | python3 runner.py weekly |
| 周期分析 | Sat 10:30 | python3 market_cycle.py --send |
| 全市场扫描 | Manual on-demand | python3 runner.py screen |
All jobs auto-skip weekends and Chinese public holidays via TradingCalendar.
The screener is manual on-demand (3-5 min runtime). To schedule it weekly,
use openclaw cron add to register a weekend slot.
| Data | Source |
|---|---|
| Real-time Quotes | Sina Finance API |
| Daily K-lines (A-share) | Baostock |
| Daily K-lines (HK) | AKShare |
| Financials (latest) | Baostock |
| Financials (10yr history) | AKShare |
| GDP / PMI / CPI / M2 | AKShare (NBS) |
| Bond Yields | AKShare (10Y gov bonds) |
| Margin Trading | AKShare: stock_margin_sse |
| FX Rate (USD/CNY) | AKShare: currency_boc_sina |
| IPO Pipeline | AKShare: stock_ipo_info |
| Stock Buybacks | AKShare: stock_repurchase_em |
| Market Indices | Baostock: SH/CSI300/ChiNext |
The screener scans the configured universe (default CSI 800 + ChiNext composite, ~700 stocks after liquidity prefilter) on demand and produces two tracks of candidates:
| Track | Pattern | Use case |
|---|---|---|
| A \u2014 Bottom Accumulation | 周线底部温和放量 + 走势已启动 + 基本面反转 | Left-side, patient |
| B \u2014 New-High Momentum | 距 52 周高 \u22645% + RS\u226580 + MA \u591a\u5934\u6392\u5217 + \u91cf\u80fd\u7a81\u7834 | Right-side, momentum |
# Full scan (~5 min cold cache, ~40s warm cache)
docker exec --user node -w /home/node/.openclaw/skills/china-stock openclaw \
python3 runner.py screen
# Outputs:
# data/stock-data/reports/screen-{YYYYMMDD-HHMM}.md ← full markdown report
# data/stock-data/reports/skipped-{YYYYMMDD-HHMM}.json ← skipped breakdown
# WeChat summary (Top 10) sent automatically# View 10-year financials
docker exec -w /home/node/.openclaw/skills/china-stock openclaw python3 -c "
from stock_data import get_provider
p = get_provider()
for f in p.get_financial_history('600519'):
print(f'{f[\"year\"]} ROE:{f[\"roe\"]:.1f}% 毛利:{f[\"gross_margin\"]:.1f}% 净利增:{f[\"profit_growth\"]:+.1f}%')
p.cleanup()
"
# Manually run any cron job
docker exec -w /home/node/.openclaw/skills/china-stock openclaw python3 runner.py close
# Check cron status
docker exec openclaw cat /home/node/.openclaw/cron/jobs.json# Gateway logs
docker compose logs -f
# Stock analysis logs (daily)
docker exec openclaw cat /home/node/.openclaw/stock-data/logs/stock-$(date +%Y%m%d).log| Measure | Details |
|---|---|
| Network | Gateway listens on 127.0.0.1 only |
| Auth | Token-based authentication |
| Container | no-new-privileges, dropped NET_RAW/NET_ADMIN |
| Git | .gitignore + pre-commit hook blocks secrets |
| Secrets | data/openclaw.json, data/identity/, data/devices/, data/stock-data/notifier.json |
# Verify no secrets will be committed
git diff --cached --name-only | grep -E "data/(openclaw|identity|devices)" && echo "⚠️ SECRETS!"┌─────────────────────────────────────────────────┐
│ Cron Scheduler │
│ morning │ midday │ close │ alert │ weekly │ cycle│
└────┬─────┴───┬────┴───┬───┴───┬───┴───┬────┴──┬──┘
│ │ │ │ │ │
▼ ▼ ▼ ▼ ▼ ▼
┌─────────────────────┐ ┌──────────────────────────┐
│ runner.py │ │ market_cycle.py │
│ + TradingCalendar │ │ Howard Marks 14-dim │
└─────────┬───────────┘ └────────────┬─────────────┘
│ │
▼ ▼
┌───────────────────────────────────────────────────┐
│ analyzers.py │
│ CANSLIM │ Value │ Growth │ TechCalc │ Signals │
└─────────────────────┬─────────────────────────────┘
│
▼
┌───────────────────────────────────────────────────┐
│ stock_data.py │
│ │
│ Quotes: Sina Finance (real-time A+HK) │
│ History: Baostock (A-share daily K-lines) │
│ AKShare (HK stock daily K-lines) │
│ Finance: Baostock (A-share latest quarter) │
│ AKShare (A/HK multi-year history) │
│ │
│ Cache: file-based + in-memory, auto TTL │
└─────────────────────┬───────────────────────────┘
│
▼
┌───────────────┐
│ notifier.py │
│ WeCom webhook │
│ (auto-split) │
└───────────────┘
stock_data.py → Data layer (Sina/Tencent/Baostock/AKShare + cache)
analyzers.py → CANSLIM + Value + Growth scoring + screener helpers
report.py → Deep-dive report with 10yr financials + technicals
market_cycle.py → Howard Marks 14-dim macro analysis (real GDP/PMI/M2/Bonds)
trading.py → Multi-dimension trading signals (trend/momentum/volume/pattern)
runner.py → Cron entry + China trading calendar (auto skip holidays)
screener.py → 全市场 CANSLIM 扫描 (双路线, A=底部蓄势, B=创新高动量)
screen_report.py → 扫描报告生成 (企业微信摘要 + markdown 完整报告)
notifier.py → WeChat sender (auto-split long messages)
- A-shares: 6-digit (e.g.,
600519for 茅台) - HK stocks: prefix
hk(e.g.,hk00700for 腾讯) - Indices:
000001(上证),000300(沪深300),399006(创业板指)
- Investment Philosophy — CANSLIM, Value, Growth methodologies
- Requirements & Architecture — Full system design
- Skill Documentation — Technical details
- WeCom Setup Guide — WeChat notification setup
Private project. Not for redistribution.