Valuation Interpretation Card

Problem

When engineers, buyers, or product managers ask whether a stock is expensive, cheap, or undervalued, a common failure mode is to split PE, PB, PEG, and PR into separate metric cards and then make a conclusion from one multiple. This skill compresses the question into one judgment card: it compares price with earnings, assets, growth, and cycle position inside the same fact base, instead of treating each ratio as a standalone verdict.

How it works

The main output is a valuation mini-table with metric, value, and five-year percentile; an industry comparison table is included only as a secondary aid. The skill fetches data through the unified facade wm.sh run, then checks methodology, enrich_plan, quality, and flags, reads only required_reads, and avoids do_not_load items. Verdicts are constrained to enumerations: industrial Fair / Low / High / Insufficient data, and financial Discount / Reasonable / Premium / Insufficient data. Without a matching flags marker, it does not label diagnoses such as 'denominator expansion', 'fake cheap', or 'Schiller divergence'; it reports visible percentile numbers and facts. Before delivery, it follows verify-checklist checks and avoids treating raw JSON tables as a pricing memo.

Boundaries

Use it for valuation fit, earnings yield, cheap/expensive, and valuation-memo questions. For business-model questions, use the company business skill; for a one-page company snapshot, use the company card; for three-statement financial data, use the statements skill. It does not output buy or sell points, avoids hunter words such as 'deeply undervalued', 'bubble', or 'sweet spot', and does not promise that every deliverable is perfect.