What problem does it solve? Investors researching a new industry often miss important stocks, get distracted by hype-driven "story stocks", or lack a repeatable standard for narrowing a whole market down to a few names worth deep research. This Skill provides a disciplined four-layer funnel that filters 30-60 candidates down to 3 final picks with documented elimination reasons at every stage. ## Core Features & Use Cases - Four-Layer Funnel: Full-market scan (A-shares, HK, US, international, pre-IPO) → 5 hard value-investing criteria (PE, ROE, cash flow, debt ratio, moat) → structured 300-500 word analyses → four-master deep dives (Duan Yongping, Buffett, Munger, Li Lu) on the final 3. - Audit & Visualization Tooling: Ships a report_audit.py script that randomly samples 15% of financial data points for source verification with pass/fail verdicts, plus SVG chart generation for funnel statistics. - Bias Controls: Explicit countermeasures for AI research biases such as large-cap preference, English-source preference, and story-stock preference. - Use Case: Tell the agent "run an industry funnel on AI compute" and receive a Chinese-language Markdown report with per-layer elimination tables, three final picks with position sizing, ETF alternatives, and a data-confidence self-assessment. ## Quick Start Ask the agent to run the industry funnel screening on a sector like innovative drugs or robotics and produce the full funnel report with the final three picks.
