Trying to make autoresolve results better

[Dave Scarface]

Hi guys,

I thought I would ask this question here because I couldn't think of a better place on the whole net!
Xeryx and team still have my greatest respect for trying to push this game to its very limits.

Actually though the question I am going to ask you is more about the way autoresolve works. Now some people might think autoresolve is only important for hotseats where human v human battles are auto resolved. But when you think about it the outcomes of some AI battles in single player are greatly effected also. I am thinking units like horse archers, how is it that say an army of spearmen slaughter an army of horse archers? Shouldn't the result be the exact opposite? but of course I want it to be more balanced not the other way round. Some eastern factions like the mongols really suffer because their best men are all horsemen.

So is there a way we can add some new code in to adress this somewhere? Without increasing unit stats and making horse archers ridiculously overpowered in battle mode? My aim is to try and make autoresolve results BETTER! I dont want a standard build of spearmen to always WIN all the time. I want different army compositions to have successful auto results as well. I believe this would be of tremendous value to the hotseat community especially. Thank you for taking the time to read my post and I will pass on any helpful tips to our hotseat community.

my regards,
David

[wolf-yop]

what about adding them light_spear only 4 bonus against cavalry and giving them more defensive combat points?, they will get better against other infantry and worse against cavalry, and in battle the formation of spearmen will not overpower the cavalry.

for last hope you can raise the horse archers bow damage and giving them less arrows.

[Achilla]

Here are few problems ...

- primarily MELEE attack and TOTAL DEFENSE combined with soldier count (number of soldiers per unit) affect the auto-resolve power, this naturally discriminates against cavalry & siege engines in any form, especially horse archers with traditionally sub-par melee characteristics
- things like morale, stamina, quality of training, AP, attack delay, missile attack practically have no influence on the auto-resolve strength of the unit

If you design it for hotseat only, you have an added bonus of not caring about negative influence of AR-points boosting secondary hit points, since no real-time battles will (should) take place. Therefore, you can change secondary hit points for non-elephant units from 0 to 1, 2 and 3, each subsequent modifier acting as a multiplier of auto-resolve strength (1x,2x,3x,4x). With secondary hit points, you can make say a cavalry unit 2 times as powerful over campaign auto-resolve bar (doesn't work for custom battles).

From what I know tinkering with other stats to achieve similar effect is waste of time, auto-resolve is a simple math calculation, it doesn't simulate behaviour of individual units, it simply adds/removes points from score, adds some random +/- factor and then decides on who is the victor. With that in mind, since no real-time battles take place, it would be wiser to simply standardise each unit to the same number of men and the same stats, then divide the units into quality tiers. As far as I'm concerned, hotseat isn't about battles but interactive campaign, so it would be nice to have balanced auto-resolve results no matter the faction involved. I know it's a bit of work, but there is no other way. Otherwise, few favoured factions with spearmen/late heavy infantry spam will shred the opposition to pieces.

Adding new code? Hell no, very doubtful. The little 'code' we are left with primarily deals with responses to sieges and is buggy, which is of no use to you.

[Dave Scarface]

Thank you Achilla, in answering my question it seems you have given me a lot to ponder on.

And it looks like I may need to come out of modding retirement yet again.. lol.

Dave

[Achilla]

Thank you Achilla, in answering my question it seems you have given me a lot to ponder on.

And it looks like I may need to come out of modding retirement yet again.. lol.

Dave

Good time to 'come out of the cave', so to speak M2TW modding doesn't disappoint, mate, after all these years. Hint: http://www.twcenter.net/forums/showthread.php?t=409073

[xeryx]

Great Questions Dave! AR is very important for balance, but cannot be done with complete accuracy..it mostly ends up a feel of balance by playing with the numbers. It is unfortunate, and many have tried to alter the effects, with unit modifications without success. Thanks Achilla for answering Dave and for giving him a reason to un-retire.

It will be time for phase 3, of XAI. Once, I get settled into the new house..prepare for a new Battle AI. We are going to revamp it once again, maybe once we get these darn castles fixed we can really do more refinement.

[diadok]

GREAT !!

[Dave Scarface]

Hey everyone,

I just wanted to let you all know that a new version of KGCM has been released to the community. This is a version tailored towards hotseat balance, but includes XCAI 4.0 as well as a completely rebalanced EDU and building tech tree. The result is a much better balanced mod between factions in KGCM, although it is still not perfect. You will find the improvements to be a fantastic improvement for single player as well.

For any XAI players who wish to expereince playing KGCM with the latest XCAI here is the LINK

regards,
Dave

[Thefallenhun]

Maybe this is a stupid question and I don't want to divert from more serious discussion by asking questions I'm sure you already know the answer to, but how exctly is autoresovle calculated. I think there has to be a random throw of the dice involved, as the same autoresolve can turn out multiple ways. Also, I'm not terribly familiar with how this game is coded, but I am familiar with a few basic principles of coding, so how "open" is the coding related to AR, are there ways to affect the calculation involved in doing it, or must it involve unit tinkering?

[xeryx]

Here is the serious answer, no one really knows the exact calculations. Yes, there is a random factor thrown in. The big problem is that is not accurate enough to be reliable for all unit types. So you then have to step back and take an abstract view of the calculations. Then establish your baselines, for the different unit types. Create many campaign battles, and test and test until you get some approximate conditions. Then keep testing the same values throughout a campaign for consistent results.

It is pretty much trial and error, and this is also a mod per mod bases.

[Thefallenhun]

So the code involved in autoresolve isn't well understood? My understanding is that certain parts of the code are much better understood both based on people working with them and with creative assembly releasing them. So do we know precisely what section of the code is involved in it? Or do we lack even that?

I can't help but think this might be a problem vaguely up my alley, as I do genetics research and rummagging around in poorly understood code, making small changes, and seeing their effects is kind of what I do for a job.

[xeryx]

No, What I said was, no one knows the exact formula. We have things we can adjust and play with, to get things more balanced. XAI does a pretty fair job of getting things closer to what they should be.

[Thefallenhun]

Is it possible to alter values that affect the autoresolve outcome without changing the unit stats?

[xeryx]

yes it is and it has been done..go try KGCM Kingdoms Grand Campaign mod..you will like it.