taiwan-hidden-champion-radar

Scan and rank Taiwanese niche-market leaders with evidence scoring and bias-free backtesting.

Updated Aug 22, 2026
One-click install
npx skills add https://github.com/JustinChangTW/ai-skills-core --skill taiwan-hidden-champion-radar-justinchangtw
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: taiwan-hidden-champion-radar
Source: https://github.com/JustinChangTW/ai-skills-core/tree/main/skills/08-finance-property/taiwan-hidden-champion-radar
Command: npx skills add https://github.com/JustinChangTW/ai-skills-core --skill taiwan-hidden-champion-radar-justinchangtw

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve? Identifying Taiwanese companies with genuine global top-three or niche leadership positions is hard because market share claims are often self-reported, unverified, or distorted by survivorship and look-ahead bias. This Skill separates verified champion quality from stock valuation and validates its own scoring radar through rigorous historical backtesting. ## Core Features & Use Cases - Champion Scanning & Scoring: Evaluates listed, emerging, and unlisted Taiwanese companies across eight dimensions (niche position, moat, globalization, leading signals, switching costs, financials, cluster advantage, evidence quality) with a 100-point rubric and A–E evidence grading. - Market Share Verification: Checks numerator, denominator, market definition, and independent sources before accepting any "world number one" claim, flagging unverified assertions. - Bias-Free Backtesting: Runs point-in-time backtests using only data published before the cutoff date, computing confusion matrix, precision, recall, F1, and balanced accuracy via the included Python script. - Use Case: Ask the Skill to scan Taiwan's machinery sector for hidden champions as of a 2019 cutoff, then backtest whether the radar's scores predicted which companies achieved global top-three status by 2024. ## Quick Start Use the taiwan-hidden-champion-radar skill to scan Taiwanese companies in a chosen industry for hidden champion potential and rank candidates with evidence grades.

Frequently Asked Questions about taiwan-hidden-champion-radar

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find hidden champion companies in Taiwan?▼

Define the universe, industry, and cutoff date first, then score each company across eight dimensions like niche market position, technology moat, globalization, and leading signals. Companies are classified as proven champions, candidates, early signals, or insufficient evidence based on verified sources.

How to verify a company's world market share claim?▼

Check the product definition, numerator, denominator, data year, and source independence before accepting any ranking. Claims with only company self-declaration or no identifiable market denominator are marked unverified and capped at half the market position score.

How do I backtest a stock screening model without look-ahead bias?▼

Use only data published before the cutoff date, based on publication date rather than data period, and include failed and delisted companies to avoid survivorship bias. Lock the model on a training period, then evaluate once on a holdout period using the backtest_metrics.py script.

What metrics does the backtest script calculate?▼

The backtest_metrics.py script reads a CSV with score and actual outcome columns, then outputs the confusion matrix, precision, recall, specificity, F1, and balanced accuracy at a configurable score threshold. Results are printed as JSON.

Does this Skill provide stock buy or sell recommendations?▼

No. It outputs research candidates and evidence grades only, explicitly separating company quality from stock valuation. It does not guarantee returns or issue trading instructions; listed companies receive supplementary financial and valuation analysis instead.