Data Sources

Data Provided

• Annual reports (PDF)
• Corporate announcements
• Insider trade notices
• Substantial holder notices
• Top 20 shareholder lists

Update Frequency

Annual reports: yearly. Announcements: as filed.

Method

PDF download and structured extraction using AI (Claude Haiku). Board composition, remuneration, qualifications, attendance, auditor fees, and shareholder data are extracted from published annual reports.

Known Limitations

Extraction accuracy depends on PDF formatting. Some companies use scanned images or non-standard layouts that reduce extraction quality. Data may lag actual filings by days or weeks.

Database Tables

Data Provided

• Daily stock prices (OHLCV)
• Company officer names and roles
• Director biographies
• Gender, birth year inference
• GICS sector/industry classification
• Analyst consensus (target prices, buy/hold/sell, EPS estimates)

Update Frequency

Prices: daily (may be delayed up to 24h). Officers: periodic refresh. Consensus: updated as brokers publish.

Method

Yahoo Finance quoteSummary API (assetProfile, financialData, earningsTrend, recommendationTrend modules). SGX tickers use .SI suffix.

Known Limitations

Yahoo officer data may lag board changes by months. Biographies are often short role descriptions, not detailed profiles. Analyst consensus reflects Yahoo's aggregation and may differ from Bloomberg or Refinitiv. Some small-cap tickers have no analyst coverage.

Database Tables

Data Provided

• Board of directors (names, roles, independence, committees)
• Executive compensation (exact figures or $250K bands)
• Non-executive director fees
• Board meeting attendance
• Auditor fees and non-audit fee ratios
• Top 20 shareholders
• Director qualifications and biographies
• ESG metrics (emissions, diversity, safety)
• Related party transactions

Update Frequency

Annually, extracted after each company publishes their AR (typically Q1-Q2 of the following year).

Method

PDFs downloaded from company IR pages or SGX EDGE. Text extracted via pdfjs-dist (with password handling for protected PDFs). Structured data extracted via Claude Haiku with section-specific prompts.

Known Limitations

Currently covers 59 of 202 SGX-listed companies. Coverage is biased toward large-cap and STI constituents. Some older annual reports (pre-2022) have broken download URLs. AI extraction may misparse unusual table layouts or non-English content.

Database Tables

Data Provided

• Company UEN (Unique Entity Number)
• Registered address
• Director appointment records (partial)

Update Frequency

As available. ACRA data integration is in progress.

Method

ACRA BizFile API and public registry.

Known Limitations

Full ACRA integration is not yet complete. Director cross-referencing between SGX and ACRA is partial.

Data Provided

• Governance Risk Score — 8 dimensions, A-F grade
• Management Credibility — 6 dimensions, A-F grade
• Accounting Quality — Beneish M-Score, Piotroski F-Score, Sloan Accruals
• Dividend Safety — 8 components with S-REIT logic
• Earnings Surprise Predictor — 8-factor model, SGX-tuned weights
• Fair Value Estimation — DCF, DDM, EV/EBITDA relative
• Financial Distress — Altman Z-Score, Z''-Score, Piotroski
• Greenwashing Detection — narrative vs evidence gap
• Auditor Quality — fees, independence, rotation
• Director Skills Taxonomy — AI-classified from biographies
• Career History — regex-extracted from bios

Update Frequency

Recomputed on each page load from underlying data.

Method

Open-source algorithms with disclosed methodologies. No proprietary or black-box scoring. Full model documentation at /methodology.

Known Limitations

Models use simplified assumptions. Fair value estimates should not be used as the sole basis for investment decisions. Skills classification accuracy varies by biography quality.