Thanks to all for the replay
Rune your retreating example is very interesting and surely I will remember it for next battle I will create, anyway I'm only half in agreement with you .
Because you implicity are assuming that all your 5 AI plans are likewise good, so in short for you the rule is same troops, different AI planes, same chance of success .
According to me usually it's not so. There is ever a better AI plan for the battle maker, and the chance of success for the player in that plan are lesser than for the others plans. So the victory parameter should be different .
In this matter there are a lot of other subtle implications that you know better than me, for example programming an AI attack plan the battlemaker can assuming the attacker dont know exactly the defender's force and the position so he can choose to set up a suboptimal but more realistic attack, or can simulate a delay and so on .
In short Rune, I think with the same troops it's possible making different and interesting AI plans but usually more are different these plans more different are the chance for the player .