How the score on your screen actually gets made

The 10 legendary-investor models

Rather than inventing one proprietary formula, the engine runs each stock through ten separate models, each one built to mirror the actual documented approach of a well-known investor or strategy. A stock that looks great to a value investor and terrible to a momentum trader is a genuinely different situation than one every model agrees on — and that disagreement is useful information, not noise to average away.

• Buffett — quality and durability: consistent returns on capital, low debt, a real moat.
• Lynch — growth at a reasonable price (GARP): earnings growth relative to the multiple you're paying for it.
• Graham — deep value and balance-sheet safety, the original margin-of-safety approach.
• Greenblatt — the Magic Formula: cheap stocks with high returns on invested capital.
• CAN SLIM — earnings acceleration and institutional accumulation (the classic growth-momentum screen).
• Zweig — trend-following with a macro filter, avoiding stocks fighting the broader market.
• Minervini — technical trend template: is this stock actually in a confirmed uptrend structure.
• Piotroski — the F-Score: nine binary tests of balance-sheet and profitability improvement.
• Acquirer's Multiple — cheap relative to operating earnings, the deep-value acquisition lens.
• Quality Compounder — steady, reinvestable growth with high returns sustained over time.

Sector-relative, always

Every score ranks a stock only against its own GICS sector — a bank is never compared against a software company's margins. This matters because a single template applied identically to every industry (a common shortcut in the category) systematically misjudges capital-intensive and regulated businesses.

The composite score

The composite blends quantitative factor percentiles with a qualitative AI reading of recent filings and earnings calls — roughly 60% quantitative, 40% qualitative. The quant side is deterministic and reproducible from the same inputs; the qualitative side is explicitly labelled as AI-generated synthesis, not fact, everywhere it appears.

Fair value: a real 2-stage DCF

Fair value isn't a single sell-side price target passed through — it's an independent 2-stage discounted cash flow model with bear/base/bull scenarios, built from the company's own reported financials. The "Price / Fair Value" ratio on the screener is exactly the number the model produces, shown identically to every visitor regardless of tier.

The data-freshness audit

Every report carries an audit of when its underlying inputs were actually last refreshed — prices, filings, estimates, ownership data — rather than a single "last updated" timestamp that could paper over one stale input. A number that hasn't been genuinely refreshed is disclosed as such, not silently carried forward.

What this is not

None of the above is personalized investment advice, and no output here is a recommendation to buy, sell or hold any security. It's a documented, repeatable calculation applied identically to everyone who looks at the same ticker on the same day — read the full disclaimer for the exact legal framing.