Taiwan Equity Research Coverage

Taiwan Equity Research Coverage (My-TW-Coverage)

Skill by ara.so — Daily 2026 Skills collection.

A structured equity research database covering 1,735 Taiwan-listed companies (TWSE + OTC) across 99 industry sectors. Each report contains a business overview, supply chain mapping, customer/supplier relationships, and financial data — all cross-referenced through 4,900+ wikilinks forming a searchable knowledge graph.

Installation

git clone https://github.com/Timeverse/My-TW-Coverage
cd My-TW-Coverage
pip install yfinance pandas tabulate

Project Structure

My-TW-Coverage/
├── Pilot_Reports/             # 1,735 ticker reports across 99 sectors
│   ├── Semiconductors/        # 155 tickers
│   ├── Electronic Components/ # 267 tickers
│   ├── Computer Hardware/     # 114 tickers
│   └── ... (99 sector folders)
├── scripts/
│   ├── utils.py               # Shared utilities
│   ├── add_ticker.py          # Generate new ticker reports
│   ├── update_financials.py   # Refresh financial tables + valuation
│   ├── update_valuation.py    # Refresh valuation multiples only (fast)
│   ├── update_enrichment.py   # Update business descriptions from JSON
│   ├── audit_batch.py         # Quality auditing
│   ├── discover.py            # Buzzword → related companies search
│   ├── build_wikilink_index.py# Rebuild WIKILINKS.md index
│   ├── build_themes.py        # Generate thematic investment screens
│   └── build_network.py       # Generate D3.js network graph
├── WIKILINKS.md               # Auto-generated browsable wikilink index
├── network/index.html         # Interactive D3.js wikilink network
├── themes/                    # Thematic investment screens (auto-generated)
└── task.md                    # Batch definitions and progress tracking

Report Format

Each ticker report is a markdown file at Pilot_Reports/<Sector>/<TICKER>_<Name>.md:

# 2330 - [[台積電]]

## 業務簡介
**板塊:** Technology
**產業:** Semiconductors
**市值:** 47,326,857 百萬台幣
**企業價值:** 44,978,990 百萬台幣

台積電為全球最大晶圓代工廠，專注於 [[CoWoS]]、[[3奈米]] 先進製程...

## 供應鏈位置
**上游:** [[ASML]], [[Applied Materials]], [[SUMCO]]
**中游:** **台積電** (晶圓代工)
**下游:** [[Apple]], [[NVIDIA]], [[AMD]], [[Broadcom]]

## 主要客戶及供應商
### 主要客戶
- [[Apple]], [[NVIDIA]], [[AMD]], [[Qualcomm]]

### 主要供應商
- [[ASML]], [[Tokyo Electron]], [[Shin-Etsu]]

## 財務概況
### 估值指標
| P/E (TTM) | Forward P/E | P/S (TTM) | P/B | EV/EBITDA |
|-----------|-------------|-----------|-----|-----------|
| 28.5      | 22.1        | 9.3       | 7.2 | 16.4      |

### 年度財務數據
[Annual 3-year financial table with 14 metrics]

### 季度財務數據
[Quarterly 4-quarter financial table]

Key Commands

Add a New Ticker

# Basic (auto-detect sector)
python scripts/add_ticker.py 2330 台積電

# With explicit sector
python scripts/add_ticker.py 2330 台積電 --sector Semiconductors

Update Financial Data

# Single ticker
python scripts/update_financials.py 2330

# Multiple tickers
python scripts/update_financials.py 2330 2454 3034

# By batch number (see task.md for batch definitions)
python scripts/update_financials.py --batch 101

# By sector
python scripts/update_financials.py --sector Semiconductors

# All 1,735 tickers (slow)
python scripts/update_financials.py

Update Valuation Only (~3x Faster)

Refreshes only P/E, Forward P/E, P/S, P/B, EV/EBITDA, and stock price — skips full financial statement re-fetch.

python scripts/update_valuation.py 2330
python scripts/update_valuation.py --batch 101
python scripts/update_valuation.py --sector Semiconductors
python scripts/update_valuation.py                          # All tickers

Discover Companies by Buzzword

Find every Taiwan-listed company related to a theme or technology:

# Basic search
python scripts/discover.py "液冷散熱"

# Auto-detect relevant sectors (skips banks/insurance/real estate for tech terms)
python scripts/discover.py "液冷散熱" --smart

# Tag matching companies with [[wikilinks]] in their reports
python scripts/discover.py "液冷散熱" --apply

# Apply + rebuild themes + rebuild network graph
python scripts/discover.py "液冷散熱" --apply --rebuild

# Limit to a specific sector
python scripts/discover.py "液冷散熱" --sector Semiconductors

Common buzzword examples:

• "CoWoS" — TSMC advanced packaging supply chain
• "HBM" — High Bandwidth Memory ecosystem
• "電動車" — EV component suppliers
• "AI 伺服器" — AI server supply chain (148 companies)
• "光阻液" — Photoresist suppliers and consumers
• "碳化矽" — Silicon carbide (SiC) companies

Update Enrichment Content (Bulk AI Research)

Prepare a JSON file, then apply to specific tickers, batches, or sectors:

python scripts/update_enrichment.py --data enrichment.json 2330
python scripts/update_enrichment.py --data enrichment.json --batch 101
python scripts/update_enrichment.py --data enrichment.json --sector Semiconductors

Enrichment JSON format:

{
  "2330": {
    "desc": "台積電為全球最大晶圓代工廠，專注於 [[CoWoS]]、[[3奈米]] 先進製程，為 [[Apple]]、[[NVIDIA]] 等科技巨頭提供晶片製造服務。",
    "supply_chain": "**上游:**\n- [[ASML]] (EUV 微影設備)\n- [[Applied Materials]] (薄膜沉積)\n**中游:**\n- **台積電** (晶圓代工)\n**下游:**\n- [[Apple]]\n- [[NVIDIA]]",
    "cust": "### 主要客戶\n- [[Apple]] (約25%營收)\n- [[NVIDIA]]\n- [[AMD]]\n\n### 主要供應商\n- [[ASML]]\n- [[Tokyo Electron]]"
  },
  "2454": {
    "desc": "...",
    "supply_chain": "...",
    "cust": "..."
  }
}

Audit Report Quality

# Single batch
python scripts/audit_batch.py 101 -v

# All batches
python scripts/audit_batch.py --all -v

Audit checks:

• Minimum 8 wikilinks per report
• No generic terms in brackets (e.g. [[公司]], [[產品]])
• No placeholder text remaining
• No English text in Chinese-language sections
• Metadata completeness (板塊, 產業, 市值, 企業價值)
• Section depth (業務簡介, 供應鏈位置, 主要客戶及供應商, 財務概況 all present)

Rebuild Wikilink Index

python scripts/build_wikilink_index.py

Regenerates WIKILINKS.md — a browsable index of all 4,900+ wikilinks categorized as Technologies, Materials, Applications, and Companies. Run after any enrichment update.

Generate Thematic Investment Screens

# Build all 20 themes
python scripts/build_themes.py

# Single theme
python scripts/build_themes.py "CoWoS"

# List available themes
python scripts/build_themes.py --list

Output in themes/ — each page shows companies grouped by upstream/midstream/downstream role.

Generate Interactive Network Graph

# Default: min 5 co-occurrences
python scripts/build_network.py

# Fewer edges for cleaner view
python scripts/build_network.py --min-weight 10

# Only top 200 nodes
python scripts/build_network.py --top 200

Opens network/index.html in browser — D3.js force-directed graph. Node colors:

• 🔴 Red = Taiwan company
• 🔵 Blue = International company
• 🟢 Green = Technology
• 🟠 Orange = Material
• 🟣 Purple = Application

Wikilink Graph — Core Feature

The wikilink graph is what makes this database powerful. Every [[entity]] in every report creates edges in a knowledge graph.

Search by entity to find related companies:

Browse: Open WIKILINKS.md for the full categorized index.

Code Examples

Read and Parse a Report

import re
from pathlib import Path

def get_report(ticker: str, reports_dir: str = "Pilot_Reports") -> dict:
    """Find and parse a ticker report."""
    base = Path(reports_dir)
    # Find the file across all sector subdirectories