Arg, my post on .com has disappeared. Ill write it out again.

Alrighty, the theory about second unit is correct in most parts. It doesnt influence victory though. That is based on army size. Anyways, the best strategy for using autocalc is to set up best troops in 2nd slot and descending order. Do NOT put missile troops in early slots as they will take all the casualties as well as any kills they score. Missile troops take more casualties anyways.

My premarius army is 4 v, 6 h, 6 p, and 3 t.

My setup used to be velites/hastati/principes/triarii in unit cards. However, after my research, I changed it to

hastati/principes/triarii/velites.

Note: I said use best quality troops, however there are times you want to use low tier troops to soak up damage because you can replace them in small towns, whereas high tier troops have to be sent to cities.

As for experience, I think there is a bug where non combative units get lvls way too fast using autocalc. My theory revolves around the "bonus kill/soldier at end of battle if odds were decent" (CA statement). My theory is that there is another repeating trigger bug, like command stars, where the bonus is repeatedly tested and given to each soldier.

As for the combat units that take casualties, I believe the kills are not distributed until losses are taken away. Thus that 80-90 men left will get all the kills. In a real battle, you notice that most of the men who do the killing are the ones who die as well.