love-stock

Standalone stock-analysis agent toolkit for China A-shares, US stocks, and Hong Kong stocks.

This repository now contains one recommended deep-wide analysis skill plus multiple standalone strategy skills. The recommended entry point is love_stock_deep_wide/: it collects stock data, recent candles, chip data where available, news, market-wide context, sector/breadth data, then produces a clean Markdown report suitable for email delivery.

Author: Miraclemin

中文说明见 README.zh-CN.md.

Strategy Learning Page

Open strategy-learning.html in a browser to learn all built-in strategies visually. The page includes a Chinese/English language selector and three simple SVG examples for each strategy: when to buy, when to wait, and when not to enter.

Recommended Skill

love_stock_deep_wide/

Use this for day-to-day analysis. The final report contains:

1. Market-wide recap
2. Core conclusion
3. Quote and data status
4. Technical-method table
5. Fundamentals and catalysts
6. Risks
7. Suggestions for empty-position users
8. Suggestions for holders
9. Data confidence
10. Interface-status appendix

Quick run:

cd love_stock_deep_wide
python scripts/analyze.py --check-config
python scripts/analyze.py --stocks 600487 --no-email

Send a finalized current-agent report:

python scripts/analyze.py \
  --send-final-report /path/to/final_report.md \
  --base-report ~/.stock_analysis/reports/report_xxx.md

What The Deep-Wide Report Does

• Uses Tushare first for stock and market data, then falls back to Tencent, AkShare, or Yahoo when needed.
• Adds a market-wide recap before individual-stock analysis.
• Includes recent 10 daily candles in the prompt, so candlestick patterns such as One Yang Three Yin can be judged directly.
• Outputs a technical-method table that compares trend, pullback, breakout, MA cross, range, sentiment, theme, event, growth, Chan theory, wave theory, and other methods.
• Produces a clean email version: main report first, interface status as an appendix.

Market-Wide Recap Logic

For A-shares:

• Index trend: Shanghai Composite, Shenzhen Component, ChiNext alignment, volume, support/resistance.
• Fund sentiment: up/down breadth, limit-up/limit-down structure, turnover, high-position divergence.
• Main sectors: leading/lagging sectors, event catalysts, leader stocks, downside spread.

For Hong Kong stocks:

• Hang Seng Index, Hang Seng Tech, HSCEI trend.
• Southbound flow and liquidity/risk appetite.
• Tech/platform, financial, property, consumer, and defensive rotation.

For US stocks:

• SPX, NDX, DJI alignment.
• Treasury yields, USD, inflation/Fed narrative.
• AI, semiconductor, software, megacap tech, financials, energy, defensives, and small-cap rotation.

The market module answers:

• Should I trade tomorrow?
• Which directions should I watch?
• Which stock signals are more credible?
• What conditional plan should I follow tomorrow?
• What conditions prevent impulsive open-market decisions?

Strategy Skills

love_stock_deep_wide already includes the analysis logic represented by the standalone strategy skills and combines them into one broad comparison table inside the final report. If you only want one method, each stock_*_analysis/ directory can still be executed independently with its own SKILL.md, scripts, configuration template, and reporting references.

Directory	Strategy	Principle	Best For
love_stock_deep_wide	Deep-wide analysis	Market recap plus broad technical-method comparison	Default complete report
stock_bull_trend_analysis	Bull trend	MA5/MA10/MA20 alignment, rising MA20, controlled pullback, volume breakout	Trend-following stocks
stock_shrink_pullback_analysis	Shrink-volume pullback	Uptrend pullback to MA5/MA10 with contracting volume	Low-risk trend entries
stock_volume_breakout_analysis	Volume breakout	Resistance breakout confirmed by volume above 2x short average	Platform breakouts
stock_ma_golden_cross_analysis	MA golden cross	MA5 crossing MA10, MA10 crossing MA20, MACD/volume confirmation	Early trend starts
stock_bottom_volume_analysis	Bottom volume surge	Deep drawdown followed by high-volume bullish stabilization	Reversal attempts
stock_box_oscillation_analysis	Box range trading	Buy near range support, reduce near resistance, confirm breakouts by volume	Range trading
stock_one_yang_three_yin_analysis	One Yang Three Yin	Big bullish candle, three shrinking small/bearish candles, then bullish breakout	Short-term pattern confirmation
stock_dragon_head_analysis	Dragon head	Sector leadership, relative strength, active turnover, catalyst resonance	Sector/theme leaders
stock_hot_theme_analysis	Hot theme	Theme strength, stock relevance, relative strength, retreat risk	Theme rotation
stock_emotion_cycle_analysis	Sentiment cycle	Turnover, volume pulses, news tone, MA compression, volatility	Emotion bottom/top detection
stock_event_driven_analysis	Event-driven	Earnings, policy, order/product, M&A, regulatory event impact path	News/announcement catalysts
stock_expectation_repricing_analysis	Expectation repricing	Whether new information improves, realizes, or invalidates market expectations	Earnings/valuation rerating
stock_growth_quality_analysis	Growth quality	Revenue, profit, cash flow, ROE, valuation durability	Medium/long-term growth stocks
stock_chan_theory_analysis	Chan theory	Fractal, stroke, segment, central range, divergence, buy/sell point	Structural trading
stock_wave_theory_analysis	Elliott wave	Five-wave impulse and three-wave correction with Fibonacci levels	Swing structure analysis

Run a single strategy:

cd stock_volume_breakout_analysis
python scripts/analyze.py --stocks 600487 --no-email

Data Source Priority

Data	Priority
A-share daily bars	Tushare daily
A-share quote supplement	Tushare realtime plus Tencent fields
A-share chip data	Tushare cyq_chips -> cyq_perf -> AkShare/Eastmoney -> local estimate
A-share market indices	Tushare index_daily -> Tencent indices
A-share market breadth	Tushare daily full-market statistics -> AkShare
A-share sectors	Tushare ths_index + ths_daily -> Tushare moneyflow_ind_ths -> Tushare moneyflow_ind_dc -> AkShare
HK daily bars	Tushare hk_daily -> Yahoo
HK market context	Tushare HK data -> Yahoo index/ETF proxies
US daily bars	Tushare us_daily -> Yahoo
US market context	Tushare US data -> Yahoo index/ETF proxies
News	Tavily / SerpAPI / SearXNG

Installation

git clone https://github.com/Miraclemin/love-stock.git
cd love-stock/love_stock_deep_wide
python -m pip install -r requirements.txt
python scripts/analyze.py --check-config

The first config check creates:

~/.stock_analysis/.env

Edit it, then check again:

python scripts/analyze.py --check-config

Configuration

Configuration priority:

process environment variables > ~/.stock_analysis/.env > local .env > defaults

Common fields:

TUSHARE_TOKEN=
TAVILY_API_KEYS=
SERPAPI_API_KEYS=
SEARXNG_BASE_URLS=
SEARXNG_PUBLIC_INSTANCES_ENABLED=true

NEWS_MAX_AGE_DAYS=3
REPORT_LANGUAGE=zh

EMAIL_ENABLED=false
EMAIL_SMTP_HOST=smtp.gmail.com
EMAIL_SMTP_PORT=587
EMAIL_USE_TLS=true
EMAIL_USER=
EMAIL_SENDER=
EMAIL_PASSWORD=
EMAIL_FROM=
EMAIL_TO=

Full setup guide: love_stock_deep_wide/references/configuration.md.

Usage

# A-share
python scripts/analyze.py --stocks 600487 --no-email

# US stock
python scripts/analyze.py --stocks NVDA --no-email

# Hong Kong stock
python scripts/analyze.py --stocks 00700.HK --no-email

# Mixed markets
python scripts/analyze.py --stocks 600487,NVDA,00700.HK --no-email

# Chinese names require Tushare
python scripts/analyze.py --stocks 北方华创,亨通光电 --no-email

Skill Usage

Put the desired skill directory under your client skill directory. For daily complete analysis:

~/.codex/skills/love_stock_deep_wide/

For strategy-only analysis:

~/.codex/skills/stock_volume_breakout_analysis/
~/.codex/skills/stock_shrink_pullback_analysis/

Reports And Email

Data collection writes:

~/.stock_analysis/reports/report_YYYYMMDD_HHMMSS.md
~/.stock_analysis/prompts/prompt_YYYYMMDD_HHMMSS_1_CODE.md

After the current Agent writes the final report:

python scripts/analyze.py \
  --send-final-report final_report.md \
  --base-report ~/.stock_analysis/reports/report_YYYYMMDD_HHMMSS.md

The email body uses the final report as the main content and appends interface status at the end.

Data And Safety Notes

• API keys are stored in ~/.stock_analysis/.env; do not commit that file.
• --check-config masks secrets as set, set(n), or missing.
• Missing optional providers degrade the report; the tool should not fabricate data.
• This project is for research and workflow automation only. It is not financial advice.

License

MIT. See LICENSE.