Every chart here answers one question from the same SEC Form 4 stream the rest of the site reads, and states the sample it was drawn from. Where the data cannot answer a question, it says so instead of drawing something.
$17,011,351 of open-market buying across 12 Aug 2026 to 26 Aug 2026. Bar height is dollar value reported that day, not a running total.
Counts, not values — one $2M purchase and one $2K purchase are both a single bar unit here. Direction only; neither colour means good or bad.
Median purchase $44,908. Bands are fixed and stated rather than quantiles, so the same data always tells the same story.
Median 1 day from transaction to filing, longest 12. 113 (18.8%) took more than two days. These are calendar days — the SEC's two-day rule counts business days, so a weekend can push a compliant filing over.
Roles are parsed from the free-text title on the Form 4 and are mutually exclusive. A title matching nothing known stays Unclassified rather than being forced into a bucket.
Companies where two or more different insiders each reported an open-market purchase inside the window. Clustered buying is uncommon, so a short list is a real result.
The distribution of the model's published tiers across cases currently open. A tier ranks how unusual a purchase looks against its own filing history — it is not a return forecast. How the tiers are built →
Two analyses in the original brief are absent. Both would need something this data cannot honestly provide, and an invented chart is worse than a missing one.