Skip to content

What (One-liner)

Calculates company business model quality from yfinance three statements + description (9 metrics including CapEx / ROIC / FCF / R&D / 3y CAGR) + qualitative tags, outputs B score 0-100. B-dimension input for BLCV composite score.

How to use

Drag blcv_business_v1 into Canvas, fill in symbol. Outputs 9 quantitative metrics + qualitative_tags (heuristic classification: light/heavy asset, high/low R&D, stable/volatile FCF, etc.) + b_score.

Core formulas

B_score = (quantitative_score + qualitative_tag_bonus) / (max_pts) × 100

Key metrics:

  • CapEx Intensity = capex / revenue (lower is better, generally < 10% counts as asset_light)
  • ROIC = NOPAT / invested_capital (>15% counts as high_quality)
  • Revenue CAGR (3y) (stable growth counts as bonus)
  • R&D Intensity = rd / revenue (tech industry > 10% counts as innovation_heavy)

Assumptions & applicability

Assumptions: qualitative tag rules are valid for current business model classification; 3-year window represents steady state; quantitative metric thresholds target mature industries.

Applicability: Listed companies in developed markets, entities such as manufacturing / consumer / tech / pharma.

Out of scope: Loss-making companies / early-stage startups / financial institutions / undergoing M&A restructuring.

Input / Output contract

symbol: str{b_score, metrics: dict, qualitative_tags: list, logic_lineage: list}. See frontmatter for details.

Known limitations

  1. qualitative_tags rely on keyword matching → yfinance description drift will affect it
  2. 3-year window insufficient for cyclical industries (energy / materials)
  3. CapEx threshold 10% is empirical, not robust across industries
  4. Adding new metrics will break max_pts normalization

References

Buffett Shareholder Letters (moat) + Porter (1985) Competitive Advantage + Damodaran (2015) Narrative and Numbers.

Golden Test

tests/golden/fixtures/tier1/blcv_business/, integration_regression (AAPL yfinance snapshot), 1e-6 tolerance, 4 tests.

Changelog

  • 1.1.0 (2026-04-20) — 0-100 scale alignment + lineage sync
  • 1.0.0 (2026-04-19) — Initial production (0-1)

Verifiable intelligence for the decisions that demand scrutiny.