You’ve gotta use division instead of subtraction or you just can’t compare different sizes sensibly. We know Knives Out, TGS, Spiderverse has great legs and BvS poor ones, but in post OW cash BvS will look much better.
Also, record breaking OWs often have very good classical legs (total/OW) and even below new #1 records I don’t think there is (or at least, was) the kind of robust negative correlation between big OWs and weak multis that you seem to be implying.
However, I’m definitely not saying that total/OW should be the one true objective measure of holding quality. It’s ultimately rather arbitrary and probably most used only because it’s convenient (in other words, lazy 😛 ). True FSS legs are better imo and I use them a lot — helps adjust for different preview sizes some. Total/first 7 days is realistically much better than classical legs calculation — including weekdays automatically helps balance for school vs summer vs winter, handle holiday Mondays, and handle midweek releases. (Total-previews)/(true first 7 days) even better yet.
And if you wanted to penalize films less for opening big then you could try perhaps post-1stSun/(first 7 days post 1stSun), or post-ThPost1stSun/(first 7 days post ThPost1stSun), some decreasing weighted mix of first week multi, 2nd week multi, etc.