Stock Analysis

Configurable stock analysis pipeline producing a Decision Dashboard
(buy / hold / sell).

Technical → Intel → Risk → Strategy → Decision

Zero configuration — free, keyless data sources (Tencent Finance, AkShare, yfinance).

Prerequisites

• Python 3.10+
• uv (dependencies auto-install on first run)

Architecture

tools/       ← bottom layer
scripts/     ← upper layer (CLI wrappers, depends on tools/)
strategies/  ← YAML strategy definitions
references/  ← detailed docs loaded on demand

Path Convention

• SD = <this skill directory>/scripts
• TD = <this skill directory>/tools

---

Usage

Run Analysis

uv run "SD/pipeline.py" analyze TICKER

Pass a 6-digit code for A-shares (601919) or a ticker symbol for US stocks (AAPL).

Pipeline Modes

uv run "SD/pipeline.py" analyze TICKER --mode MODE
uv run "SD/pipeline.py" modes

Mode	Stages	Use Case
full	Technical → Intel → Risk → Strategy	Complete analysis (default)
quick	Technical → Strategy	Fast signal, skip news/risk
news	Intel → Risk	Sentiment & risk screening only
technical	Technical	Indicators only

Output Keys

Key	Content
technical	MA, RSI, volume ratio, trend score, support/resistance, K-line pattern
intel	News headlines, risk alerts, positive catalysts
risk	Risk categories, PE/PB valuation
strategy	Detected market regime, up to 3 recommended strategies
dashboard_schema	Decision Dashboard JSON schema

Keys only appear for stages included in the selected mode.

Other Commands

uv run "SD/pipeline.py" schema    # Print Decision Dashboard JSON schema

---

LLM Task

After calling pipeline.py analyze, interpret ALL returned data and produce
the final answer as a formatted Markdown report.

Signal weighting (without strategies): Technical 40 %, Intel 30 %, Risk 30 %.
With strategies: Technical 30 %, Intel 25 %, Risk 25 %, Strategy 20 %.

Scoring: 80–100 buy (high conviction) / 60–79 buy / 40–59 hold / 20–39 sell / 0–19 sell (major risk).

Risk veto: If any risk flag has severity "high" or veto_buy: true, cap the final signal at "hold".

Strategy evaluation: For each recommended strategy, check entry conditions
against the technical data and adjust the score accordingly.

Read references/output-template.md for the full Decision Dashboard
Markdown template with all placeholders.

---

Tools

Five tools under tools/, each runnable standalone via uv run "TD/<name>.py":

Tool	Purpose
get_daily_history	Historical OHLCV K-line data
get_realtime_quote	Real-time stock quote
get_technical_indicators	MA, RSI, support/resistance, trend status
search_stock_news	News with impact classification
is_stock_hot	Check if stock is leading in hot sectors

Read references/tools-reference.md for CLI usage, return schemas, and
the internal dependency graph.

---

Strategies

11 built-in strategies in strategies/, auto-selected by market regime:

Regime	Strategies
sector_hot	dragon_head, emotion_cycle
trending_up	bull_trend, volume_breakout, ma_golden_cross
trending_down	shrink_pullback, bottom_volume
sideways	box_oscillation, shrink_pullback
volatile	chan_theory, wave_theory

Custom strategies: drop a .yaml into custom_strategies/ — auto-loaded, same-name overrides built-in.

Read references/strategies-guide.md for regime detection logic,
strategy evaluation flow, required_tools, and the full YAML template.