How AIR Scores a Filing: Our Methodology, Step by Step

A research tool that won't explain its method is asking you to take it on faith. We'd rather show the work. This is what happens, every night, between a company filing a disclosure and a scored event appearing on AIR the next morning.

Step 1: The nightly read

After the US market closes, our engine collects the day's SEC filings across the universe we cover — roughly 1,800 small- and mid-cap US companies. The focus is on event-driven disclosures, 8-K filings above all, because that's where material changes are announced first.

The engine doesn't skim headlines or wait for press coverage. It reads the filing text itself — the body and what it discloses — the same source a careful human researcher would start from.

Step 2: Classification

Each filing is classified by what it actually is. For 8-Ks that starts with the item numbers — a material agreement (Item 1.01), a completed acquisition (Item 2.01), an officer change (Item 5.02) — and continues into the substance: what kind of event, in what part of the business.

Classification matters because the same form carries very different news. "An 8-K was filed" is close to meaningless; "an 8-K disclosing a completed acquisition" is the beginning of a research question.

Step 3: Materiality, weighed consistently

Here is the core of the methodology — and the part that only works because it's consistent. The engine weighs how significant each event looks from the filing itself: the nature of the event, the specifics disclosed, and how much the text says it affects the business.

The output is a score. Not a prediction, not a recommendation — a structured, repeatable judgment of how much this event deserves a researcher's attention.

One lens, applied the same way across every company, every night. A human team of any size can't do this evenly; judgment drifts between analysts and between Tuesdays. An engine applies the same standards to the 1,800th company as to the first.

Step 4: Publication — and the part we can't edit

Scored events are published to the Daily Radar and to each company's Timeline, free to read. And each month-end, our track record locks in the five widest Discovery Gaps as a timestamped public entry that we never edit afterward. The methodology above produces the scores; the record is how you can judge, over time, whether that methodology is worth your attention.

What the scores are not

A score is a reading aid, not advice. It tells you an event happened, what kind, and how significant it looks within our framework — so your research time goes where it's most likely to matter. What to conclude, and what to do, stays where it belongs: with you.

AIR is a research tool, not investment advice. We surface and score public information; decisions are yours.