Why generic insider buying is a noisy signal
Form 4 fires for every officer trade above any threshold. 10b5-1 plan trades, option exercises, automated quarterly lots — most of the volume isn't a discretionary buy decision. If you treat all Form 4 buys equally, you're rewarding the average S&P 500 mid-level VP for executing a calendar trade they signed up for two years ago.
The selection bias is brutal. Insiders sell ~10x more often than they buy, but most of the selling is option-grant-funded tax cover, not bearish. Most of the buying, paradoxically, is also non-discretionary (10b5-1 buys, ESPP). Filtering this down to the discretionary-buy tail is where the signal lives.
The CEO+CFO same-period cluster filter
The strongest insider read on the EDGAR feed is: CEO and CFO buy stock in the same 14-day window, both above $250K, neither under a 10b5-1 plan. That filter fires roughly 20–40 times per quarter across the entire US-listed universe. Forward returns over the next 90 days, on a market-cap-weighted basis, run ~12–18% above benchmark in our backtests.
Why does it work? Both senior officers reach the same view independently and act on it materially. The information asymmetry is concentrated. The 14-day clustering rules out calendar-driven 10b5-1 prints (those are spaced quarterly). The $250K floor rules out symbolic 'show of confidence' buys that don't change either officer's net worth.
Add a third rail — same-period independent director buy — and the hit rate climbs again. The full /insiders/ ranking weights all three plus officer tenure (higher conviction for officers in seat > 18 months).
What we ignore: VP buys, 10b5-1 prints, ESPP
VP-level Form 4s have weak forward returns. Why: information asymmetry at the VP level is genuine but narrow. They know their division. They don't know the consolidated outlook. We log VP buys for completeness but score them at one-fifth the weight of CEO/CFO clusters.
10b5-1 plan buys (footnoted on the Form 4 filing itself) get auto-filtered. Same for ESPP and option exercises. The /insiders/ table flags them with a muted icon so you can verify the filter is working.
What we add: tenure, prior-buy track record, sizing context
We tag each insider with their prior-buy track record. An officer whose last three discretionary buys ran +20% on average scores higher than a first-time buyer. The track record is public per officer; the score adjustment is published in the /methodology/ page.
Sizing context: a $500K buy from a CEO with $50M in vested stock is sized at 1% of net worth — meaningful but not extraordinary. A $500K buy from a CEO with $2M in vested stock is sized at 25% — extraordinary. The /insiders/ ranker normalizes by net worth, not absolute dollars.
How to use /insiders/ alongside the catalyst feed
Workflow: open /insiders/ daily. The top 5 are the discretionary-cluster reads — typically 1–3 fire on any given day. Cross-check against /scanner/. If the ticker is also showing a catalyst on /scanner/ in the past 14 days, the convergence is the entry. If it's clean of catalysts, it's a slow-roll position.
/preview/ includes the daily /insiders/ top 3 in the 4 AM ET email. The full ranked list is on the /insiders/ page itself, free tier capped at 3 rows, Reader at $9 unlocks the rest.
Get the same picks our paying readers get — free
Top 3 of /scanner/ and /jackpot/, the audited /trust/ log, and the daily catalyst email. No drip campaigns, no upsell sequences — one note per trading day.