In this example the attacker will sustain 0.1 (10%) of original losses, while defender will lose 7 times more troops, than normally.

With these numbers the human would win automatically pretty much all autoresolve battles,

That's great. Do you know what the 0.9 and 0.1 in front mean? Win Chance...90% but how can that be if vastly outnumbered. It just increases the win chance 90% from the current value?

One more thing: How can I set up a log to receive these messages? And why does it appear like you have an extra "end" at the wrap-up at the bottom?

Yeah, the values are 0-1 so 1 means 100% win chance.

What log are you speaking of?

out.ting as in out.ting("defender is major"). I'm testing it with human added in. It works but not as expected. I seem to lose just as badly when at a disadvantage, but my barely-win predictions usually turn into major wins. I wonder if the 7 number is out of range as well.

I switched it around and put human at disadvantage. It appears that there is no difference. In fact, some tests showed better results where this script was reversed (calling for a loss). Can we talk about the source of this mod? What situation was the script developed for? I'm testing it on Veteran battle difficulty and hard campaign settings mostly, but have also had similar results on hard hard and veteran veteran.

How your script looks like?

I have established a baseline by loading the game without the script but with campaign var autoresolver set to minimal:
autoresolver_battle_result_accuracy_high_prediction_variation 0.0010
autoresolver_battle_result_accuracy_low_prediction_variation 0.0010
autoresolver_battle_result_accuracy_moderate_prediction_variation 0.0010
autoresolver_battle_result_accuracy_very_high_prediction_variation 0.0010
autoresolver_battle_result_accuracy_very_low_prediction_variation 0.0010
battle_combat_outcome_prediction_non_artillery_missile_accuracy 0.0010

The script add on is at the end:
if attacker_is_major == true and defender_is_major == false then
--out.ting("major attacker v minor defender")
cm:modify_next_autoresolve_battle(0.9, 0.1, 0.1, 7, false) --attacker win chance, defender win chance, attacker losses modifier, defender losses modifier
elseif attacker_is_major == false and defender_is_major == true then
--out.ting("minor attacker v major defender")
cm:modify_next_autoresolve_battle(0.1, 0.9, 7, 0.1, false)
elseif attacking_faction:is_human() == false and defender_is_major == false then
--out.ting("minor attacker v major defender")
cm:modify_next_autoresolve_battle(0.7, 0.1, 0.1, 7, false)
end
end
end,
true
)
end

If you can advise, I can help test next weekend. I would really like this to work. Right now, I am seeing most results about the same as not using the mod. The autoresolver seems to throw in highly variable results in each direction about 25% of the time.

Any suggestions for that script? I have 2 lines where the human is defender that follow the pattern of the first one I added. I wonder if there are too many "end"s in there and it is not even running. I've tried it with all kinds of number values even .1, .9., 7,.1 as attacker doesn't budge the result average.

Is your script for Troy or other TW?

That script is just an add-on to the one you posted here for this mod. I'm still waiting to know if it can be written better or the numbers have limits or something to make it work.