Update on the insider cluster tracker ..turned it into something a few people asked for, here's what changed

Posted a few weeks ago asking if tracking insider buying clusters on

small caps was actually useful or if I was overthinking it ([EARLIER POST](https://www.reddit.com/r/pennystocks/s/eQiHgdVbPS)) . Appreciate everyone who weighed in,

35upvotes and a bunch of genuinely useful comments pushed me to actually

finish this instead of letting it sit as a half-built script.

Quick recap for anyone who missed it:

it flags when 2+ insiders buy

the same stock within about a week of each other, filters out scheduled 10b5-1 trades (not real signal), and only looks at small/micro caps under \~$2B market cap.

A few things came out of the comments that changed how I built it:

\- Someone pointed out most existing insider trackers (OpenInsider, etc.) just list every individual filing

— you still have to manually spot when multiple insiders are buying together. So I added actual cluster detection instead of a raw feed ... it groups buys by ticker and time window automatically instead of you scrolling through hundreds of rows.

\- A few people asked about scoring/ranking — added a basic model that weights insider count, dollar amount, and whether it's an officer (CEO/CFO) vs just a director, since a couple comments made the case that officer buying is a stronger signal than a 10%owner buying.

\- Someone asked if I track how these actually perform after — so now I'm keeping a running log and plan to post honest results (including the ones that don't work out) instead of just cherry-picking wins.

How that would be if I Turned it into a free weekly email since a couple people in the last thread asked to keep seeing this ongoing rather than one-off posts ..

I think smart people already knows the pattern, that penny stocks turned more fortunes than mid or large cap did.

happy to drop the thing for anyone who wants it, no pressure either way. Still mostly just want to know if this is actually predictive or if I'm pattern-matching noise, so if anyone wants to poke holes in the

methodology I'm all ears.