Define the universe
Choose liquidity, geography, sector, size, and reporting-history requirements before looking at individual names. A stable universe reduces hindsight and makes comparison more honest.
Use multiple lenses
Business quality, earnings quality, financial health, revisions, catalysts, market trend, momentum, narrative, and valuation answer different questions. Keep the inputs and weights visible when combining them.
Apply fatal-flaw gates
A strong average should not hide an essential failure. Missing filings, severe liquidity pressure, unreliable financials, rights-blocked data, or an unmodeled obligation may require an unavailable or held state.
Keep long-term and near-term views separate
A durable business view and a near-term technical or catalyst view can disagree. Do not silently blend them into one score. Name the horizon and methodology for each view.
Compare valuation assumptions
A lower multiple is not automatically cheaper, and a higher multiple is not automatically expensive. Compare growth, margins, capital needs, durability, dilution, and scenario assumptions.
Make the next evidence visible
A useful shortlist includes the question, current evidence, risks, missing inputs, and the next event that can change the read. That makes the process reviewable even when the future outcome differs from the thesis.