How the shortlist is built

Methodology

A transparent quantitative screen for finding companies worth deeper, qualitative research.

What the screen does

Each edition starts with the current S&P 500 universe and looks for companies whose reported profitability, cash generation, growth and valuation make them worth further research. It is a comparative screen, not a fair-value model or a recommendation to buy.

Financials and real-estate companies are excluded because their economics require sector-specific measures that are not part of this general-purpose model. A maximum of two companies from one sector may appear in an edition.

Eligibility rules

  • A complete valuation calculation using the required SEC and price inputs.
  • Positive median free cash flow and net income across the measured five-year window.
  • Positive free cash flow in at least four reported fiscal years.
  • No duplicate SEC company identifiers in the final shortlist.

Attractiveness score

The 0–100 score combines seven disclosed components. The site displays it on a 0–10 scale for readability.

Free-cash-flow yield25%
Earnings yield20%
Enterprise value / operating income15%
Operating margin10%
Five-year annualized revenue growth10%
Positive free-cash-flow consistency10%
Net debt / free cash flow10%

Valuation and operating-margin percentiles are calculated within a sector when at least five eligible companies are available; smaller groups use the global eligible-company distribution.

Quality score

Quality is separate from shortlist order. It combines operating-margin percentile (40%), positive free-cash-flow consistency (35%) and revenue-growth percentile (25%). Leverage is discussed separately because debt-tag coverage in SEC filings can be incomplete.

Data and publication review

Company fundamentals come from the SEC Company Facts API. Approximate valuation prices use a median of up to five recent daily closes from HF Data Library. The date beside each company shows the exact price and filing coverage used.

Research notes are generated deterministically from those inputs. Numerical warning rules flag possible anomalies, but they cannot identify every accounting, competitive, regulatory or management risk. Each edition must pass a human publication review before deployment.