Created a skill based on the IPO scoring criteria mentioned in the article and uploaded it to GitHub. It's already backtested and calibrated using real Hong Kong IPO data from 2025-2026. Works pretty well in opencode, but feels a bit overthought in Claude code. Features: 1. Eight-dimensional scoring system (max score: 100): subscription popularity, industry track, sponsor, cornerstone investors, greenshoe, valuation, market cap, base score 2. VETO one-vote rejection rule: 2 data-driven VETO rules + retail frenzy exemption mechanism 3. Intelligent text parsing: automatically extracts key fields like sponsor, cornerstone, subscription multiples from search text 4. Historical backtest validation: 127 real Hong Kong IPOs from 2025-2026 Repo: lianyanshe-ai/hkipo