Method

How Pegasus Alpha works

One ticker in, a full living report out. The rules below hold for every report, and no setting can weaken them.

The product in one paragraph

Pegasus Alpha is a research tool for finance firms. It writes deep, cited equity research and keeps it up to date. It is not a stock picker: it gives no rating and no price target. Where a report shows a value range, that range is model output, not a target.

Rules that never bend

  1. No fake data. Every number, node and link traces to a stored row or a cited source. A number remembered by a model is not a source. Unknown stays unknown.
  2. Evidence tiers on every claim. Each claim is a fact, an inference, SIMULATED or ML, and carries a tier from 1 to 5. Anything below tier 3 is treated as unconfirmed. Simulations are never evidence.
  3. Point in time. A fact is dated by when it became knowable, not by the event date. Nothing after the report date leaks into a report.
  4. Free and public data. Reports are built from public sources such as SEC filings, market data and macro series. The sources are listed in each report.
  5. Gaps are shown. When data is missing the report says what is missing and why, instead of filling the space.

What happens on a run

  1. The engine gathers data for the ticker and writes a numbered evidence pack.
  2. It maps knock-on effects through the Web: chains from suppliers, customers, regulators and events to the company.
  3. Claims are written into a living research core for that ticker, each with its evidence, tag, tier and as-of date.
  4. Reports and other views are rendered from the core. Prose may only use claims that exist in the core.
  5. When the world changes, only the touched claims are re-checked and a change set records what moved.
See it on real output: read a full sample report, check the public accuracy record, or read how agents and Excel connect.