Detect insider buy clusters before the news cycle

When 3 or more insiders at the same company file Form 4 buys inside 24 hours, that’s a cluster. Single insider buys are noisy; clusters tend to precede positive catalysts. insider_buy_clusters does the grouping for you.

Tool

insider_buy_clusters. params:

param	default	notes
`limit`	5 (free) / 25 (pro)	max items
`min_filers`	3	minimum cluster size
`buys_only`	true	exclude sell clusters

Agent prompt

Call insider_buy_clusters with min_filers=3 and buys_only=true. For each
cluster, output: ticker, filer_count, total_usd. Then call sec8k_material_today
with the same tickers to see if there's a recent 8-K. Comment on whether
the cluster appears to be reaction or anticipation.

Expected output

ticker  | filers | total_usd  | 8-K?
ABCD    | 5      | $2.1M      | 5.02 officer change (3 days prior) -> reaction
WXYZ    | 4      | $890k      | no recent 8-K -> anticipation
QRST    | 3      | $1.4M      | 2.02 earnings (2 days prior) -> post-earnings buy

What to do next

The pure “anticipation” rows are the alpha. Reactive buys after good news are weaker.
Cross-check with confluence_today to see if the ticker is showing up in other signals.
Position sizing: total_usd is the dollar value of all insider buys in the cluster, not a price target.

Limits

Free tier returns top 5, max 10. Pro returns 25 default, 50 max. Cache: free 24h, pro realtime.