How to read an insider signal
Insiders file every trade with the SEC on Form 4. Most filings are routine. These five questions — the same filters our score applies automatically — separate a trade worth your attention from noise. None of them makes a trade predictive on its own; together they tell you how much weight it deserves.
1. Buy or sell?
The two are not mirror images. Insiders sell for a hundred reasons — taxes, diversification, a house — but there is essentially one reason to spend their own cash buying their own stock on the open market. A buy carries more information per dollar than a sell of the same size. Heavy selling still matters, especially when it clusters (question 2) or lands near an event (question 4).
2. One insider, or a cluster?
One director buying is an opinion. Three or more insiders independently trading the same direction inside a month is the single most-watched pattern in this data — separate people with separate information reaching the same conclusion. We count independent decisions, not co-signers on one filing, and we tag each member's role: a cluster of executives reads very differently from affiliated funds rebalancing together. Current clusters appear at the bottom of Trending.
3. Scheduled plan, or a discretionary trade?
Many insider trades execute automatically under Rule 10b5-1 plans scheduled months in advance — they say nothing about what the insider thinks today. Every number on this site already excludes them (we check both the filing's checkbox and its footnotes, because the checkbox under-reports). If you read filings elsewhere, this is the first thing to check.
4. Is earnings close?
A discretionary trade in the weeks before a results announcement is the most informative kind — it is also when insiders are usually barred from trading by company blackout windows, which makes the exceptions notable. Cards on Trending show each company's estimated next earnings date, projected from its own SEC filing cadence — treat it as an estimate, never a promise.
5. Big for this company?
$2M of buying is a shout at a company with a $500M public float and a rounding error at a mega-cap. That's why cards show the signal as a share of the float ("0.04% of float") rather than letting a raw dollar figure impress you. Always size the trade against the company, not against your own wallet.
Then: what does the prediction market believe?
Everything above is public information — the question is whether the market has priced it. That's the site's whole premise: when insiders lean one way and the related prediction market leans the other, one of them is wrong, and that divergence is worth a closer look. We surface the pattern with a transparent score, link every figure to its SEC filing, and claim nothing about anyone's intent.
Not investment advice. Form 4 data lags trades by up to two business days; check the receipts on every number before acting on anything.