Well, for one thing, I'd be careful not to overestimate the actual tactical use of amphibiousness, especially when it comes to Soviet vehicles - they're broadly amphibious by design, but that comes with some serious restrictions that'd have to be accounted for in game. And, I'd argue that by generally omitting it, the game is already doing a better job than it might seem.
BMPs especially can only very dubiously considered amphibious - even when working as such, the BMP-1 and 2 have very little reserve floatation, no tolerance for any sort of sea state or even fairly mild river current, a speed of 4mph on water at the absolute most, very little control once on water, and (most notably) no ability to fire while on water (this was corrected in the BMP-3, which was designed from the outset to be able to fire while on water - but it's not in AB, last I checked). And, like many other amphibious fighting vehicles over the years, in actual service most of them soon lost that amphibious ability altogether, trading it for better protection. Things were a bit better with BTRs and other lighter vehicles with better floatation and proper water propulsion/steering systems, but those too came with their limitations.
Few things illustrate the issue better than the fact that in Chechnya (which I've been looking at closely for my mod), the 131st motorized brigade alone lost at least 4 vehicles (flooded, stuck, and/or overturned) while crossing the Neftyanka "river" (a 3-5m stream in a gully) near a broken bridge on the morning of Dec.31st, 1994.
So I'd actually make the argument that Russian amphibious capability maybe should be treated as an operational capability rather than tactical - that is, with enough preparation, they can deploy their vehicles to a staging point across water, and then fight from there. But spontaneously deciding to go waterborne for tactical reasons would probably be really stretching it. So, unless the limitations and risks of making such a move modeled...
Well, there are already obstacles that can cause loss/immobilisation of vehicles so I doubt it would be a big problem to implement crossing losses.
The main problem I can think about is the AI. I guess there would need to be another formation/pathfinding mode for water crossing or something.
People shouldn't ask themselves why schools get shoot up.
They should ask themselves why people who finish schools burned out due to mobbing aren't receiving high enough compensations to not seek vengeance.