how we price · the machine

No black box. Here is the machine.

Every verdict on this site is the same pipeline, run on ~1,500 stocks every weekday. It is deliberately strict, openly audited, and it publishes its own misses.

Read the filings

SEC documents, not headlines. Multi-year cash flows from EDGAR, normalized so one lucky year can't become a permanent annuity. Incomplete facts stop the pipeline right here — 'unknown' never becomes 'cheap'.

Score the business

Returns on capital, margins, debt, consistency — quality before cheapness, because a cheap deteriorating business can stay cheap all the way to zero.

Value it conservatively

Three scenarios — if things go wrong, the most-likely case, if things go right. Growth is tied to what the business actually reinvests; sector routing keeps bank math away from banks it doesn't fit. A re-derived v8 model runs in shadow next to this one — October's receipts judge both.

Let the risk officer veto

Accounting forensics (earnings-manipulation scores, 'do not rely on our financials' filings, late filings) can block a buy outright — no matter how cheap, no matter how loved.

Wait

The verdict names a price, and then nothing happens — usually for months. The price decides, not us. Doing nothing on purpose is the strategy.

What we never do

No price targets from vibes — every number traces to a filing and a formula.
No AI-invented numbers — language models argue and summarize; only the math decides values.
No rewards for trading more — nothing here earns anything from your activity.
Nobody pays us to like a stock — no tickers are sponsored, ever.
No editing history — every call is timestamped; the journal is append-only.

The track-record contract

Tracking since

Jun 22, 2026

Every call

timestamped

First resolved receipts

Oct 2026

We publish the misses too — the NO-ledger counts every refusal against a coin-flip control, the disagreement board scores us against Wall Street, and the graveyard buries the signals that died. An unedited journal is the product.

The limits, out loud

A conservative model calls most great companies too expensive. We miss winners on purpose — that is the trade.

When the gap gets absurd, we say so instead of pretending: names whose market price sits far beyond anything our math can defend go to the too-hard pile — a refusal, not a target. And when an external audit argues the math itself is too harsh, the challenger runs in public shadow until the data settles it. Strictness is a choice; hiding it would be a lie.

See it working: check any stock · the live proof

Research & education only — not investment advice. Past results don't predict the future, and neither do we.