How the Conviction Screen Works

Two groups of people have unusually good information about a company. Its own officers, who run it, and large institutional investors, who research it professionally and move real money. They do not coordinate. They file different forms, on different schedules, to disclose what they did. The conviction screen watches for the rare quarter in which both groups moved toward the same stock at the same time.

What has to happen, exactly

A name appears on the screen only when both legs fire on their own strict terms:

• The insider leg: two or more insiders of the company bought in the open market within a rolling window. Purchases under pre-scheduled 10b5-1 plans are excluded, option grants and exercises are excluded, and a single large owner buying alone does not count. Independent wallets, own money, same conclusion.
• The institutional leg: among the 500 largest managers filing quarterly holdings with the SEC, the number holding the stock rose from one quarter to the next, and the net dollars they held in it rose too. Both conditions together, so a price rally alone cannot fake it.

Each row shows the evidence behind both legs: how many insiders bought and for how much, how many large holders the name gained, and the change in cohort dollars. Nothing is scored, ranked or recommended. The screen reports a coincidence of filings, and the reader decides what it is worth.

Why co-occurrence instead of either signal alone

Because the failure modes differ. Insiders can be wrong about the market even when they are right about the business. Institutions herd, and a rising holder count sometimes just means a stock joined an index. But the two groups fail differently, watch different things, and file independently. When both move the same direction in the same quarter, at least two unrelated sets of informed people reached the same conclusion with money attached. That is rarer and harder to fake than either signal alone.

The honest limits

• Both legs arrive with lags: insider filings within two business days, quarterly holdings filings up to 45 days after quarter end. The screen can only be as current as the slower filing.
• Our insider history begins in 2026, so the combined signal has closed very few measurable events. Its own base rate stays unpublished until at least 30 events exist, and the screen says so on its face. The institutional leg's base rate, computed over thousands of events since 2020, is published on the evidence pages.
• Co-occurrence is information, not endorsement. Some of the most confident agreement in market history preceded losses. The screen tells you where informed groups agreed, never whether they were right.

Where to find it

Members find the conviction screen under Research, institutions. The names, the insider dollars, the holder changes and the base-rate context all sit on one page, and every underlying filing is one click away. Educational analysis of published data. Insight, not advice.