How this platform reaches every number.

What this platform is for

Insider Scanner organises and analyses reported insider transactions so they can be investigated. The question it exists to answer is “what are corporate insiders doing, where is activity concentrating, and which filings deserve a closer look?” — not “what should I buy?”

Nothing here is investment advice, and no figure on this site should be read as a forecast. Insiders buy and sell for many reasons that have nothing to do with their view of the company: diversification, tax planning, scheduled trading plans, liquidity needs.

Where the data comes from

Three sources, each with a distinct role. Nothing on the site is hand-entered, and no figure is a placeholder.

SEC EDGAR Form 4

Open-market purchases and sales filed market-wide, synced hourly. Grants, option exercises, gifts and other non-open-market transaction codes are excluded before the file is written.

Scored research feed

Filings that cleared the research model's own bar, with the rating and evidence-strength tag the model assigned. Synced from the model's published output.

Historical evaluation

Backtest results over roughly two decades of past filings, used only on performance pages and never presented as a forward-looking claim.

Form 4 stream currently holds 600 transactions through 2026-09-18. 2 filings currently carry a research rating.

Derived metrics, and the exact rule behind each

These are computed by this application from the source data above. Each states the exact rule that produces it, so any number on the site can be traced to a definition.

Insider role

The Form 4 'title' field is free text. It is matched against a fixed, ordered list of patterns and assigned to exactly one bucket: CEO, CFO, Chair, President, 10% owner, Director, or Other officer. Order matters and is deliberate — a 'Chairman, President and CEO' is counted once, as CEO, so buckets stay mutually exclusive and always sum to the filing count. A title matching no pattern is reported as Unclassified rather than forced into a bucket.

Cluster activity

A company where two or more DISTINCT insider names reported open-market purchases inside the selected window. Counting is by distinct name, so one insider filing three times is not a cluster. C-suite participation within a cluster uses the same role patterns described above.

Weighted insider cost basis

Share-weighted mean purchase price: the sum of (price × shares) over reported open-market purchases, divided by total shares purchased. Sales are excluded entirely — averaging a sale into an 'insider cost' would describe a price nobody paid. Shown with the full minimum-maximum range of prices paid, never as a single number alone.

Period-over-period change

Compares the selected window against the window of equal length immediately preceding it. When the feed holds no prior window, the change is omitted rather than estimated, and the page says so.

Buy / sell ratio

Total reported dollar value of open-market purchases divided by total reported dollar value of open-market sales, in the same window. Above 1.00 means more money was spent buying than was realised selling. It is a description of reported flow, not a forecast.

Role buckets in use: CEO, CFO, Chair, President, 10% owner, Director, Other officer, Unclassified.

The research rating

Filings that clear the research model's bar carry a letter rating (currently A, B in the live feed) and an evidence-strength tag. The rating is a prioritisation tool: it ranks reported activity by how unusual it is, so limited research time can go to the filings most worth examining.

The rating and evidence-strength values are produced by the research model and published into this application's feed; this website displays them and does not recompute them. Where a filing carries no rating, no rating is shown — a blank is not a low score.

Known limitations

Stated here rather than left to be discovered. These analyses are absent because the underlying field is not present in the data, and deriving them would mean inventing values rather than observing them.

Sector & industry

The Form 4 stream carries no sector or industry classification, so sector heatmaps, sector rankings and industry drill-downs cannot be computed from observed data.

Ownership change

Shares held before and after a transaction are not in the feed, so percentage-ownership change and its distribution cannot be derived.

Peer comparison

Peer groups depend on sector classification, which is unavailable.

Geographic concentration

Available on the globe, by US state only. Headquarters come from SEC's own filer addresses, which give a city and a state-or-country code, so a mark means "an issuer headquartered in this state" and never a street address. Issuers outside the United States are counted but not placed.

Insider network

Insider names are free text, not stable identifiers. Treating two identical names as one person would assert a relationship the data does not establish.

Two further limits apply to everything on the site. Form 4s may be filed up to two business days after a transaction, so every figure reflects reported activity, not live activity. And insider names arrive as free text, so spelling variants of one person may count separately.

Historical evaluation

Backtest figures shown on performance pages describe how the research model's past selections behaved over roughly two decades of filings. They are a record of what happened, measured after the fact.