GS 2.3.0

GS 2.3.0 is a GS-style single-stock research workflow for A-share, Hong Kong, and U.S. equities.

This release combines the best parts of two repo lines:

• the cleaner public skills/ packaging, README structure, scaffold helper, and pipeline wrapper from gs-2.1
• the deeper v2.2.0 research discipline from gs-single-stock-pipeline-v2
• a new hard identity gate that prevents ETF / LOF / fund / index products from being forced into fake single-stock valuation
• script compatibility for both older dotted tickers such as 300124.SZ / 06656.HK and Tencent-style symbols such as sh600519 / hk00700

This repo is designed for practical equity research, not generic memo-writing. A completed GS run should force:

1. identity confirmation and live-ish market snapshot
2. deep research first
3. separation of business truth vs market-driving narrative
4. competing-narrative generation before the final thesis
5. explicit narrative maturity classification
6. mandatory fragility audit before valuation
7. valuation-router selection
8. target price / upside-downside / final stance
9. PDF-first publishing when a polished deliverable is useful

What is in this repo?

skills/
  gs/SKILL.md                         # shortcut alias skill
  gs-single-stock-pipeline-v2/
    SKILL.md                          # full v2.3.0 workflow skill
    templates/output-template.md      # report template
scripts/
  quote_snapshot.py                   # live-ish quote snapshot helper
  scaffold_input.py                   # create starter research input JSON
  build_report.py                     # render markdown report from template + JSON
  render_pdf.py                       # markdown-to-PDF with optional system backends
  run_pipeline.py                     # convenience wrapper
examples/
  sample-input.json                   # lightweight Tencent example
  a-share-industrial-sample.json      # fuller A-share valuation example
requirements.txt                      # no required Python packages
VERSION
README.md
LICENSE
.gitignore

What GS 2.3.0 Does

A completed GS 2.3.0 run should answer:

• What is this company really?
• What is the market trading right now?
• Is the provided code actually a company, or an ETF / fund / index product?
• What narrative is driving the stock now, and is it different from business truth?
• Is the narrative already in the financials, becoming material, or still optionality?
• What fragility can break the thesis?
• What valuation method actually fits this business?
• What target price follows from the assumptions?

This repo is most useful when you want repeatable single-name analysis rather than ad hoc one-off writeups.

How to Use

Option A - Use the repo as workflow documentation

Read:

• skills/gs/SKILL.md
• skills/gs-single-stock-pipeline-v2/SKILL.md
• skills/gs-single-stock-pipeline-v2/templates/output-template.md

This gives you the current workflow logic, output standard, evidence discipline, and publishable report structure.

Option B - Use the helper scripts

The core helper scripts use only the Python standard library.

1. Get a quote snapshot

python scripts/quote_snapshot.py --ticker 300124.SZ
python scripts/quote_snapshot.py --ticker 06656.HK
python scripts/quote_snapshot.py --ticker hk00700 --market hk
python scripts/quote_snapshot.py --ticker sh600519 --market cn

2. Scaffold a research input file

python scripts/scaffold_input.py \
  --company 腾讯控股 \
  --ticker 00700.HK \
  --market HK \
  --price 462.6 \
  --out examples/tencent.json

3. Build a markdown report

python scripts/build_report.py \
  --input examples/sample-input.json \
  --out output/report.md

The older --output flag is also supported:

python scripts/build_report.py \
  --input examples/sample-input.json \
  --output output/report.md

4. Render a PDF

python scripts/render_pdf.py \
  --input output/report.md \
  --output output/report.pdf

If pandoc + weasyprint or pandoc + wkhtmltopdf exists on the machine, the script uses that richer backend. Otherwise it falls back to the bundled dependency-free PDF renderer.

5. Convenience runner

python scripts/run_pipeline.py scaffold --company 阿里巴巴 --ticker 09988.HK --market HK --price 132.8 --out examples/alibaba.json
python scripts/run_pipeline.py build --input examples/alibaba.json --out output/alibaba.md
python scripts/run_pipeline.py pdf --input output/alibaba.md --output output/alibaba.pdf

Workflow Logic

The core GS 2.3.0 order is:

1. Confirm identity and live market snapshot
2. Reject fake single-stock analysis for ETFs, LOFs, funds, or indexes
3. Deep research the business first
4. Separate business truth from market-driving narrative
5. Generate competing narratives
6. Classify narrative maturity
7. Run the fragility audit
8. Choose the right valuation router
9. Show analyst-style derivation
10. End with target price and stance

Supported valuation routers

• simple_core_pe
• segment_sotp_pe
• segment_sotp_ps
• hybrid_sotp_pe_ps

Mandatory narrative split

Every analysis should separate:

• business_truth_summary
• market_driving_narrative_summary
• why_they_differ

Mandatory fragility audit

At minimum, check:

• geographic concentration
• customer / channel concentration
• supply-chain dependency
• litigation / IP risk
• regulatory / policy risk
• inventory / delivery / warranty risk

When to Use This Workflow

Use GS 2.3.0 for prompts like:

• 帮我分析 XXX
• 看看 XXX 值不值得买
• 给我做一下 XXX 的估值
• /gs 腾讯控股
• /gs 贵州茅台 重新估值

What This Repo Is Not

This is not:

• a fully automated research agent with browsing built in
• a market data vendor SDK
• a guarantee of correct valuation assumptions
• a substitute for company filings, calls, or judgment

It is a workflow package that makes single-stock research more disciplined, auditable, and shareable.

License

MIT