Well that's a correlation in and of itself... what you're telling me is there is no statistical effect at all from form 4's. To me that would mean don't bother selling or buying on them just ignore them- hahahah
Oh two other variables I forgot was the box in the form for Direct vs Indirect ownership and/or if the form contains text 105-b. If it is part of a blind trust trading plan on a D, or indirect ownership on an A my instinct is that it would have less an effect. So eliminating those forms may produce a population set with more of a correlation.
Maybe Form 3's move the needle would be another interesting check?