I was trying to devise a formula for STR and DET at fixed WD (and specifically, the only WD that matters for BiS), and while it does appear it should be a*STR*DET + b*STR + c*DET + C due to how linear DET and STR seem to be in isolation, actually trying to solve for that has given me some iffy results (notice how those lines don't intersect near the same point numbers-wise even though it looks like they should eyeball-wise). Probably just needs tons more data points, so no WD testing for now. And really, doubling all this effort for a single different WD? aaaagggg.

In a practical sense, though, consider DET for DRG at i130 to be more like .45 than .33, but I don't actually know what. Calculating stat weights from actual damage seems reasonable if you don't mind that there may be significant error from a small number of samples.