The problem is that your data is largely irrelevant to the in-game performance of the tanks and weather or not perk prices need to be changed.
Damage on penetration is perhaps the least important aspect of combat. And on top of that your testing procedure, while consistent and repeatable, does NOT take real-world circumstances into account, and thus is almost useless from a practical standpoint.
If you were advising a noob how to play, you would have given him endless frustration as you tell him his T-34/85 has a great gun, and yet he consistently struggles with penetrating enemies that aren't so stupid as to park square on into incoming rounds, and doesn't understand why they're just going through his armor like butter.
You want useful data? Compare ranges of immunity vs other tanks and their overlap (if there are any). Compare probability of first round hit at unknown ranges, compare repeatability of point of aim.
THATS what matters for fighting a tank.