Possible outcomes of a 1:1 (attacker/defender losses) prediction considering the +-1 rule: 0:0, 1:0, 0:1, 1:1, 2:1, 2:2, 2:0, 0:1 and 0:2.
There are also additional restrictions to possible outcomes: you can't obviously gain health points during combats nor destroy more enemy healthpoints than existing. That's the reason why attackers generally suffer more casualties than predicted.
Other implications are the losses/gains of xp for the unit itself and its HQ. While units gain xp by fighting, HQs have to "win" a fight to gain xp, otherwise they will lose some.
The only exception to the combat rule is in case the defending unit is fleeing during combat, usually resulting in higher damage to the defender and/or fewer damage to the attacker.
Especially in WWI-Breakthrough with the limited attaches to an HQ, this could result in losing xp and units where you never expected to; the reason I demanded to limit possible outcomes to + or -1 in the upcoming WW1 game.
Funfact: while playing WiE the randomisation has been canceled accidentally for more than half a year, and no player complained about, neither the Axis nor the Allies.
Allthough I think some randomisation is okay, I guess the reason why it`s so annoying in WaW and WiE is the generally low combat results of units like corps due to their low attack value. If the prediction would be 4:4, +-1 wouldn't have such an annoying influence. And the Axis would actually be in danger to lose some units during early russian counterattacks.