This may be of limited interest, but after posting a bunch of AI vs AI games on the ATD2 forum, I decided to give it a whirl for this game too.
In the first game, I set both sides to computer+ (the suggested setting against humans, and the Soviets got an ovewhelming victory. The last German unit was barely hanging on to Kharkov.
So I raised the Germans to computer ++ (and left the Soviets to computer +), and this time the GERMANS got an overwhelming victory . Although I did not watch either game, this seems to indicate that given the big difference in victory points obtained respectively by the Germans and the Soviets, a fairly small difference in the gameplay (such as a breakthrough at a critical point) can make a significant difference in victory points, although the results on the ground may not look that big. For example, taking heavy casualties as the Germans is much more harmful than it is for the Soviets.This assumes that two battles are statistically significant, which I believe they are, but you are welcome to disagree.
As in the case of ATD2, this also indicates that the AI is significantly better at the defence than at the attack, which is not very surprising for a computer game.This MAY indicate that against the AI, you have to be a better player to win as the Germans than as the Soviets, but there may be other considerations such as the computer ability to easily calculate complex odds versus ability to plan maneuvers to exploit enemy weaknesses.
And based onmy experiments with ATD2 (see that forum for the results), giving one side an advantage by means of increasing the comouter advantage (say like computer ++) makes a bigger difference than doing it by increasing the die roll (say by +1). The latter made no significant difference in the case of ATD2.