Strategic Combat in CoM

[Seravy]

Any way to get caster rating added to the wizard casting skill, instead of trying to get it into strategic combat directly?

That would make the problem worse instead of helping.
We want the AI to recognize the unit being more powerful than it appears.

(A = AI rating, N = Normal combat power of unit, S = Strategic combat power of units)

Ideally we want A=N=S.

A=S<N or A=S>N are acceptable - they result in the same problem, the AI underrates one side for normal combat. The AI can make up for that through smart use of combat spells (and/or higher combat skill), or if they underrated their own side, at worst they hold back and attack when they have an even stronger stack. Not optimal but we can live with that.

In case of Efreet we have this case : A=S<N.

Your suggestion would make it S=N<A - the unit would be equally good in both types of battles, but the AI would consistently underrate it still. This is fine if the Efreet is theirs, but doesn't improve the existing problem of not attacking when it's possible. If the Efreet is an enemy unit, this result in the AI sending insufficient armies against them consistently, which is not acceptable.

Long story short, my primary goal is to have A=S. Having A=N and S=N are secondary goals.

Note that "figure" mechanics are the same thing as armor - aside from armor, it makes no difference if the attack is coming from one or multiple figures.
Modifying the rating using the existing input parameters will not allow more accurate simulation of armor, no matter the formula, as none of those parameters are a function of armor, or have a strong correlation with it.
New input parameters are not possible.
If we know something can't be done in theory, there is no point trying to find one anyway in practice.
You are still trying to get the impossible done. That's pointless.
...so if you want a new formula anywhere, it should be in the defense rating where armor is an actual input.

PS : If anything, the AI not being able to consider the effect of wizard casting skill in strategic combat is already bad enough. (fortunately, the effect is minor compared to the most important stacks the AI is supposed to have at a given casting skill)

[zitro1987]

I didn't know there was an 'ai rating'. Is there a different algorithm for it? Was it shared somewhere on this thread? I did a keyword 'ai rating' search in forum and got no results.

[Seravy]

There is only one rating. The AI uses that, so I called it AI rating.

A = THE unit rating, used by the AI.
N = How useful the unit actually is in normal combat (not an actual number)
S = How useful the unit actually is in strategic combat (not an actual number)

A is calculated from the defense, attack and ranged ratings. As strategic combat uses those values, A already represents S well enough.

This topic is about making S more closely resemble N. My point is, I want A to remain an accurate representation of S so we can't change S without changing A the same way.

[zitro1987]

What do you think of tweaking hydra's armor up and hp down to reduce its overall strategic rating (while maintaining its tactical durability)?
What about reducing cost of swordsmen? The weak unit generally doesn't deserve 30 base cost, and only performs decently at elite.

[Nelphine]

Yes my caster idea isn't great. I'm OK with that. That was a single paragraph of my post, unrelated to the rest of the post.

When I say change formula, I mean things like figures +0.5. See if that gets better results, compared to an ideal rating list, than the current. I don't think it does, but without an ideal list, I can't tell, because we've shown time and again that your idea of how good given units should be is quite different to mine.

This is also why the second city troop entry needs to be level 3 - its not just about what's easiest available, its about comparing to what is still fairly easily available. That's why it's an ideal list. Its all good to show you've balanced the system for level 2 units, but if it simply breaks at level 3 that isn't helpful.

This is also why I want a level 4 adamantiim entry. But I accept its rare enough its not supremely important. (I just think its easy to add to an ideal list - do you WANT level 4 adamantium jackal riders better than great drakes?)

The list you've given us just shows where they are now, and which ones you think are super bad. But really, a fire giant should be better than level 2 halberdiers. So, to me, the point of the ideal list is to see where you think the ratings SHOULD be, not just to find out which units are worst represented right now.

[Seravy]

What do you think of tweaking hydra's armor up and hp down to reduce its overall strategic rating (while maintaining its tactical durability)?
What about reducing cost of swordsmen? The weak unit generally doesn't deserve 30 base cost, and only performs decently at elite.

That would make Hydra significantly more powerful - the primary counter to Hydras (aside from armor) is Fireball/Heroic Shout/Fairy Dust/Immolation, all of which are way less effective against 7 armor.

Swordsmen are not even worth thinking about - any decent city can produce one at both 25 and 30 cost in a single turn, and they won't be used any more often even for 25.
Besides, we pushed Halberdiers to be more cost-effective than swordsmen - now improving swordsmen would put us back to square one.

[Nelphine]

Unfortunately, I can't remember where you linked/explained the tool you use to calculate those ratings. I know you include ranged combat somehow; are you able to link/explain it again please? I'd like to do my own list of the top 40 units (as calculated at level 3 alchemy), but with level 4 adamantium, just to see what the information is. (My current formula, is square root of attack*defense - but I don't know how you're calculating ranged. )

Also, given the changes to both thrown/breath (making it equal to attack instead of 50% higher), and the change to stone/death touch, shouldn't stone/death gaze also drop down to 160?

[Seravy]

In the latest version there are two files in tools.zip.

unitsrate.pas outputs the rating for all units and that's it.
unitord.pas sorts the results and includes units at 2 levels (none and lv2 alchemy).
You can easily modify this to work with any level you want by replacing

Code:

with u do begin
inc(hit);
{inc(hp);}
atk:=atk+2;
rng:=rng+2;
def:=def+2;
end;

with whichever additional stats you want for your desired level. (inc adds one. Feel free to replace it with hit:=hit+1)
hit is To Hit, HP is hearts per figure, atk is swords, rng is bows/thrown/breath, def is shieds.

You'll need borland pascal to run/compile it.
It's recommended to redirect the output into a text file as it's too long to fit on the screen.

If you want it manually, all the formulas are in the code, the one you want is ((attack+ranged)*defense)/4096

Gaze is not per (own) figure. It shouldn't be changed.

[zitro1987]

Love these tools. Fun to experiment with the 'tweaker' and these. Sometimes the math surprises me.

Berserkers Level 2, Alchemy (Barbarian) 238 (I can keep unit comparable in tactical while I reduce it in strategic)
Angel (Life) 243 Arch Angel (Life) 492 Djinn (Sorcery) 655 ... not bad until I get Hydra (Chaos) at 2029!! despite my lower HP, higher armor idea. I'm still getting used to the math

[Nelphine]

Ah, so in that, melee and range are equal strength. Simplistic. Overly simplistic? I would say it is, I don't think there's ever a case (in strategic combat) where ranged is as important as melee. In tactical, you can use ranged to take out weak units, and to soften up strong units so your first strikers can take them out before they fight back. In strategic it's all just raw damage, which makes ranged much weaker. I'd suggest something like ((attack+2*ranged/3)*defense)/4096. But overall, not a huge deal.


Odd. I'm not getting the same number as you with that formula.. my colossus is 1184, yours is 1172. My demon lord is 777, yours is 769. Bizarre. Oh I've accounted for 9G changes, that makes sense. Nope, you know what? My hydra doesn't match yours either. I dunno. Everything on my list is just slightly higher than your list. *shrug*

I'm also doing something wrong completely on chaos spawn, as I'm getting 456. But the Chaos thread is completely not up to date on what Chaos Spawn actually has, and I don't happen to have a game available with a chaos spawn. Hum, also my Wraiths are only 331, way lower than yours. What? Doom Bats are also coming in at 181, even with 7 armor - what?

Here's the list (though obviously chaos spawn is off), but only the top 41 units (only one version of war mammoth, and I've added halfling swordsmen), with city troops at level 4 with adamantium. Note, I don't think you need to necessarily worry about this list that much, this is just a 'what do you think, do city troops get out of hand, without any buffs that require spells?' type of list. (I've put a second number for my weighted ranged formula, and in brackets listed where that second number would put it.)

Hydra (Chaos) 1838
Great Drake (Chaos) 1531
Colossus (Nature) 1184 / 1009
Great Wyrm (Nature) 867
War Mammoths Level 4, Adamantium (Troll) 848
Death Knights (Death) 799
Demon Lord (Death) 778 / 696 (between jackal riders and hammerhands)
Gorgons (Nature) 716 (these stand out as spectacularly good compared to other summons. Being twice as good as a very rare, even if it's Djinn, seems odd; I actually use these summons when I don't use city troops, they can do something. It feels like they aren't designed around the same balance as other summons, and if other summons were increased to be closer to Gorgons, I think a lot of my worry about powerful city troops would go away.)
Jackal Riders Level 4, Adamantium (Gnoll) 712
Hammerhands Level 4, Adamantium (Dwarf) 664
Behemoth (Nature) 634
Sky Drake (Sorcery) 633
Arch Angel (Life) 631 + Holy Buff 2 (holy buff is included in their own rating, but this also increases everything else in the stack as well, which is pretty huge in strategic combat. Hello jackal riders! They may have been overbuffed? Then again, that's still only going to put them up around... great wyrm-colossus level. Which is right where I think a summon from a realm weak at summoning should be, so if summons were changed, per my note in Gorgons, I wouldn't change Arch Angels any further.)
Paladins Level 4, Adamantium (High Men) 580 + Holy Buff 1 (holy buff is included in their own rating, but this increases everything else; not as good as the arch angel, but still probably effectively puts them around gorgons)
Horde Level 4, Adamantium (Orc) 579
Berserkers Level 4, Alchemy (Barbarian) 557
Slingers Level 4, Adamantium (Halfling) 545 / 436 (just above beastmen halberdiers; and halflings have poor military!)
Javelineers Level 4, Adamantium (Lizardmen) 542 / 458 (between draconian and beastmen halberdiers)
War Trolls Level 4, Adamantium (Troll) 531
Nightblades Level 4, Adamantium (Dark Elf) 496
Pikemen Level 4, Adamantium (High Men) 476
Halberdiers Level 4, Adamantium (Draconian) 468
Stone Giant (Nature) 464 / 386 (between minotaurs and doom drakes)
Chaos Spawn (Chaos) 456 (I think I've definitely done something wrong here, since you had it at 332.)
Halberdiers Level 4, Adamantium (Beastmen) 434
Halberdiers Level 4, Adamantium (Dark Elf) 434 / 386 (between minotaurs and doom drakes)
Elven Lords Level 4, Adamantium (High Elf) 402
Minotaurs Level 4, Adamantium (Beastmen) 392 (this is why I never use armorer's guilds...)
Djinn (Sorcery) 390 / 331 (between wraiths and storm giants; this is extremely bad for doomstacks, given the cost involved, except windwalking is amazing, so ideally they only want one of these in each doom stack. can the ai be taught not to include multiple of these in a doomstack?)
Doom Drakes Level 4, Adamantium (Draconian) 381
Halberdiers Level 4, Adamantium (Troll) 375
Warship Level 4, Adamantium (None) 374 / 315 (just below wraiths)
Angel (Life) 373 + Holy Buff 1 (putting them up somewhere around high men pikemen probably?)
Nightmares Level 4, Adamantium (Dark Elf) 362 / 308 (just below wraiths)
Chimera (Chaos) 356
Wraiths (Death) 331 (wait, what? why did you have 450 for these? this is as bad as my chaos spawn...)
Swordsmen Level 4, Adamantium (Halfling) 326 (My cut off for an end game unit is under 300... note that SWORDSMEN are this high... heroism plus halflings is unfreakingbelievable at the start of the game...)
Storm Giant (Sorcery) 288 / 240
Efreet (Chaos) 278 / 227 (between manticores and cockatrices)
Manticores Level 4, Adamantium (Beastmen) 239
Cockatrices (Nature) 209
Doom Bat (Chaos) 181 (whoa... this is way lower than yours too. what? and that's with the increased armor from 9G)


Note: Again, I don't think you HAVE to balance anything around this list. I'm just trying to show what can exist, and warlord AI really aren't THAT uncommon; warlord with adamantium probably happens every... 5th game, even on poor minerals? It really really makes a difference. (And that rare warlord halfling with heroism... *shudder* if only the AI knew it was a walking death machine...)