RBS has already answered this, but I would like to stress how wrong this conception is projected to the past in general and to this period in particular. The composition of most armies of this period was based on several factors, but doctrine of warfare play basically no role, and the chance for selection by the general was minimal. The factors involved were in fact quite others, for examplepantherboy wrote: I can't really agree with your view on realism as you ignore certain factors. The composition of the army was based upon their doctrine of warfare and what they felt would give them the greatest advantage.
1) Money. This one was the most important single factor. Armies were based on the available monetary resources, and cheaper was always the prefered option, to the point of recruiting soldiers with no weapons, because it was prefered to have men enlisted than to have weapons in storage.
2) Social structure. The importance of the aristocratic factor in the army was almost entirely dependant on this, regardless of actual performance in battle.
3) Tradition. That is a powerful concept in all periods except maybe in our own, so that battle performance, so important in our days, was a concept of lesser importance. The social importance and tradition of any unit was more than enough to secure it a position in any army regardless of actual battle performance. That is why the typical argument "If they keep that or that type of unit is that it worked" in fact holds no depth.
4) Available troops. The deployment of forces was dependant on many concepts, most alien to their adequation to battlefield terrain. Garrison forces were always superior in numbers to field armies, and it was always tempting to generals to draw on nearby garrisons to fill their commands.
5) Finally, pitched battles were but one of the actions in war, sieges, skirmishes, raids, ambushes, nocturnal assaults...all were more numerous than pitched battles, and armies in a campaign could not be tailored for battles even if the generals tried it, there were those other operations to count with. Horse arquebussiers were once and again to be found worthless in battle, however they were invaluable in exploration, raids and campaign in general. Regardless, no general would leave them behind in a pitched battle, numbers were always a good thing to have, even if only for the morale factor.
To sum up, this is a game and we can pretend it to be a simulation, but if we go for the Historically accurate argument, leaving the AI to select the army is closer to history than leaving it to the players.