It's perhaps just my preference, but leaders' influence on battle results are a tad high (IMHO). Even brilliant generals didn't manage to make a perfect battle order in close to every single engagement. I'd prefer if more of the results of a battle was transferred to the player's ability to move troops to the engagement province (in face of attrition) rather than how lucky he was when recruiting a general.
-------------------------------------------
About cavalry: Now I almost guaranteed to win early on if I build pure cavalry armies, to the point of exploitation.
Right now cavalry is a no brainer to choose early on, because a greater mass of infantry is vastly inferior in all respects except for assaulting. Low province support limits (which disfavours armies in quantity) and high manpower cost/combat efficiency, as well as the fact infantry gets shafted in combat (which it should admittedly), contribute to this. Something needs to counterbalance this, and I'd like that to be increased money recruitment and support costs for cavalry. Historically, a knight was a lot more expensive to set up and maintain than a lowly foot-soldier, at least 5 times and possibly 10 times, which is why infantry in many cases far outnumbered cavalry in medieval battles despite inferiority.
-------------------------------------------
About cavalry: Now I almost guaranteed to win early on if I build pure cavalry armies, to the point of exploitation.