Autobattles

[ReluctantSamurai]

Increasing the base movement rate of all units helps path-finding

And it also helps in the realm of reality. The movement of a unit (land or sea) from point A to point B on the map might take a year or two, whereas in real life, said unit could make the trip in days or weeks.

It seems that reducing the HP value used in auto-resolve for certain units might help lessen the impact of unfavorable results, but...messing with a single number or stat often has unintended consequences in other areas. I believe the stat_sec file (as mentioned by Aradan) also greatly impacts auto-resolve in the way of casualties sustained, and changing those values do not affect manual battles. That post # 71 in the third link above, nails it, I think

One of the ways I dealt with chariots was to eliminate them from the game. The Celts still had them in the time frame of the game, so I let Britannia keep theirs, but Egypt lose theirs (and Egypt was poorly done to begin with, ie. The Mummy skins, and should have been historically represented as Ptolemaic Greeks). The last recorded use of them in battle in the Middle East was in Alexanders time, and the javelin was quickly making them obsolete, in any case.

there will be other factors like walls and command stars

According to Aradan [second link in post #21]:

Experience shows that auto-resolve ignores fortifications completely, so probably that's why that happened. And yes, command plays a major part in it, as does experience level.

And after reading the accounts of players who auto-resolve siege battles, it seems to be the case that the AI ignores fortifications

MasterOfNone had this partial solution:

The best you can do is set up a SiegeStarter trait that, when it detects siege battles, gives a -10 command to the attacker and a +10 command to the defender.