Methodology
How a verdict is made
Every call on this site comes out of a six-stage pipeline built from Buffett and Munger's principles, reasoning over SEC filings, the structured financials attached to them, and licensed market data — the primary sources a human analyst would use.
The pipeline
Six stages, five of them gates
A hard fail on any of the first five — a business no one can explain, no durable moat, a management red line, a broken industry, a failing scorecard — ends the analysis with a REJECT before any valuation is computed. Only when all five pass does the sixth stage run.
01
Comprehensibility
Can a non-specialist explain this business in a few plain sentences?
A business that needs deep domain expertise to understand is rejected — outside the circle of competence.
02
Moat
Is there a durable competitive advantage, and will it still be there in ten years?
Switching costs, network effects, brand pricing power, scale, regulatory franchise, patents, habit. No identifiable moat is a rejection.
03
Management
Do they allocate capital well, and do they tell the truth?
Buybacks below intrinsic value, no value-destroying M&A, stock-based comp not hidden from headline earnings, no related-party gymnastics, meaningful insider ownership. Failures here are integrity red lines.
04
Industry
Does the industry survive the next decade?
Structural cyclicality, regulatory overhangs and technological-obsolescence risk over a 10-year horizon. Deeply cyclical or structurally-threatened businesses are rejected.
05
Quantitative scorecard
Do the numbers agree with the story?
Return on invested capital, owner-earnings yield, debt against free cash flow, share-count trend, stock-based comp as a share of revenue, and margin trend — graded A through F.
06
Valuation and memo
What is the business worth, and what would we pay for it?
Intrinsic value is triangulated across earnings-power value, a growth-adjusted discounted cash flow, a reverse DCF, and industry-appropriate models for banks, insurers, utilities and REITs. Munger cross-checks follow — inversion, a lollapalooza check, a cognitive-bias audit — and then the verdict, with a concise Buffett-style memo.
Outcomes
What each verdict means
Price is at or below our buy line — at least a 15% margin of safety to central intrinsic value, widened to 30% when a qualitative concern is flagged. A quality business at a price that justifies deploying capital.
A high-quality business trading above our buy line. We wait for a pullback to a 15% discount to central intrinsic value — deeper when a qualitative concern is flagged — before it becomes a BUY.
No actionable mispricing. Either high quality but expensive, or marginal quality at a fair price. Skip without prejudice.
Outside our circle of competence. The moat trajectory or 10-year outlook can't be judged with enough confidence to value the business, so no price is low enough to compensate. We abstain rather than guess.
Failed one of the qualitative gates — comprehensibility, moat, management, industry — or the quantitative scorecard. A disqualifying defect; no valuation is produced.
Not enough data to produce a verdict: no current price, incomplete financials. We would rather return no answer than a fabricated one.
A processing error prevented a verdict from completing. The ticker is automatically re-analyzed on the next scheduled run.
Inputs
Where the facts come from
- SEC filings
- Annual, quarterly and current reports and proxy statements — 10-K, 10-Q, 8-K, DEF 14A and their foreign equivalents — including the business, risk-factor, MD&A and executive-compensation sections, plus the structured GAAP data attached to them.
- Licensed market data
- The tradable universe, daily prices and valuation ratios, standardized fundamentals, corporate actions, and specialized bank and insurance fundamentals.
Cadence
How often it is re-run
- Weekly
- A Sunday-night batch re-analyzes every ticker in the universe. Verdicts move as fundamentals and prices do.
- On a filing
- A new 10-K, 10-Q, 8-K, 20-F or 6-K re-analyzes that ticker on the following daily pass — a filing is a material event worth reading, so it jumps the weekly queue.
- Nightly
- An incremental ingest pulls updated filings, fundamentals and prices so the next analysis runs on fresh data.
Limits
What this is not
Numora is an analytical tool, not investment advice. Verdicts are produced by an automated, model-driven pipeline applying a defined rubric to public information, and they can contain errors — a mistake in the underlying data, a misjudged moat, a valuation input that is over-extrapolated. Treat every verdict as a starting point for your own research, not a buy/sell signal.
The operators of Numora may hold positions in the securities the tool flags, and may buy or sell them as a result of its analysis. The full research-only disclaimer and conflict-of-interest disclosure are on the legal & disclosures page.