They will use the +35% value given for the first 1990s DU hull package developed (and which SEPv2 most likely has already) just because it’s the only official source stating any values, not accounting that the one implemented in 2015 may, indeed, be even better…
The same way they used Leopard TVM’s values from the 1990s to model 2A7V’s 2021’s armor under the same premise of “it’s the only numbers we have”.
I love how the 1990s armor trials involving two pre-production prototypes (Leclerc prototype, Leopard TVM (2A5 prototype)) and a downgraded export package of an early iteration of a baseline model (export M1A2) are taken like some kind of Armor Bible for 2010s-2020s tanks lmao
A vehicle where the maximum Battle Rating of the vehicle among all modes is no more than 2 steps lower than the maximum Battle Rating of the vehicle in this research tree (in the tree type in a nation).
I think the problem came from 2A7V having a completely new model so damage model and armor consistency werent transferred from Leopard 2A6 or Strv 122.