You don't understand. All the factions except the player's are fighting nearly every battle in autocalc. So yes, it matters, ALOT! There is a lot to keep in mind when you do balancing.

This autocalc issue is one of the reasons Egypt is such a powerhouse. It has two oversized units. If you match up two 20 unit armies, and half of one army is filled with cheap 50% oversized units, how do you think that AI faction is going to fare vs. other AI factions? It's going to kick the snot out of 'em. That was a very strong reason for modding Egyptian Bowmen and Desert Cav to correct sizes. (I did adjust cost as appropriate comparing to similar units.)

I don't disagree about upkeep and I have adjusted it in my game. But you need to think about it carefully, especially when considering horse archer factions.

The real problem with TW series is that the population growth tech tree doesn't work all that well anymore. We really want representative armies, rather than "all elite" armies when we hit certain population levels. The rule set doesn't handle this at all.