@PaulD @pandakraut Thanks for your replies, very much appreciated!
These are very sound tactical suggestions! Unfortunately they clash with my playstyle, since I try to avoid making decisions based on foreknowledge of the historical context. That precludes things like specializing the troop mix for particular phases of the upcoming battle - each of my divisions attempts to be relatively balanced (1 battery, 1 skirmisher/cav, the rest infantry). I also avoid "mopping up" AI troops trapped at map borders, etc, because that's just an artificial construct of the game (and isn't fun).
So in Shiloh, the most I allow myself to do is hold position until it's clear that the enemy force is overwhelming, then fall back to a more defensive position. Unfortunately the game limits this since the map is severely restricted for a number of hours - an engagement is forced where 3-star 4k brigades of confederates are able to charge a marathon distance through woods and streams and rout at least one brigade, even under double flanking fire and close range artillery - the followup attack is basically unmanageable. I feel like that initial effort is unrealistic, even for the best of troops. Can only imagine what Pickett's charge would be like later on...
IIRC the main issue is when, IMO, the AI should know it can't win and should just stop attacking. E.G. in the very first battle, 2nd phase, once the confederates' fancy train is pushed back and the initial bonzai charge has failed, idk how it thinks a second attack in exactly the same manner has a chance of working since they've fewer soldiers and artillery. I saw the same issue at river crossing and logan's crossroads... repeated attacks in the same place and manner as the initial one, but with fewer troops, artillery, and cohesion. EDIT2: Some of these followup attacks are truly baffling, since only one or two brigades are sent forward, and slaughtered - this is what I was referring to in the original post.
Noted on the modifier, I'll start out with .85 for both size and experience scaling, I think. I guess that will land me somewhere between BG and MG? I'll also give that video a watch for tips, perhaps I'm putting my career points in the wrong places? I started out getting myself over the "basic" recon hump and then splitting points between medicine and army organization to slowly grow the army and retain veterans, but I've consistently got a lot of leftover manpower.
Thanks again!
EDIT: Maybe what I should be doing is playing on MG and increasing the scaling a bit?