[LH] Yithril Playthrough AI and Balance Analysis

Map: Default settings. Medium Size. 8 Players. Challenging Difficulty.

I chose to play as Yithril because I wanted to test the AI out and this meant getting into a lot of fights with them. So I planned on being a massive warmonger.

Started near a 4,3,2 and settled there. There was also a clay pit nearby, iron, and other resources nearby. As I expanded and took out nearby monsters like Butchermen my city leveled and I made it into a Fort, in order to better be aggressive early game. 

I ran into Resoln, and some other factions but Resoln was actually close to me. Being a warmonger I attacked them off the bat. I took Gedden with little to no resistance. Then marched on Hagudst.

I designed a unit called a Grinder. Basically it had a one-handed blunt weapon, leather boots, leather breastplate, leather vambraces, muscle, and fury. With this set up it suffered no encumbrance penalty and had a lot of damage. Also using berserk they could sprint across the field with a +4 damage bonus. Is wasn't rare for them to one shot enemy units.

My Grinders made short work of Resoln's militia and unarmored spear men. Even outnumbered 2 to one I won with few to no causalities. It helped that Ceresa did odd things like leave cities under siege only to get killed by a cave bear. I was also disappointed that Ceresa had no bound elementals. Since she refused to surrender I wiped her out.

I ran into a bunch of Magnar's stacks out exploring. They were made up of some truly badly designed slave units that had 9 or even 5 initiative. Not exactly scary, even in huge stacks. I decided to attack Paridan next because they are always fun to destroy. They were located below me, sorry I don't have a good picture but you can kinda see them on the mini map.

I took New Serrene with no problem but Tenfell was filled to the brim with champions and horsemen. Of course horsemen are a horrible unit. Discounting their choice of blunt weapons which is probably a known bug they still used more armor then they should as it made them encumbered. As a result even though they had horses my Grinders were faster.

I also started producing Dreadnaughts which were my custom juggernauts with Heart of Stone, Ignore Pain, and Brutal Nature. With the removal of armor types the Heart of stone trait is OP and the splash trait is UP with cleave. Dreadnaughts had base 14 armor with HoS and a Braided Belt. With magical amulets and Fortress bonuses it got even higher. Needless to say Dreadnaughts were nigh unstoppable and made short work of Horsemen. Having the improvement that gave all my units Charge helped more then a bit as well. Grinders thus could charge across the entire field and oneshot enemy units with a +7(berserk, fury, muscle, charge) attack bonus first turn.

The above picture also shows the bad strategic AI. Even with my army right beside their town the AI sends out a group of horsemen to take back that outpost, instead of maybe attacking my solo units. They also placed their sovereign alone in the middle of my guys... with no back up what so ever. This unfortunately happens all the time. The AI just does not realize that putting small groups of units or even solo champions right next to my huge armies is a bad idea. Kinda a huge deal.

Also the enemy sovereigns had multiple specializations. The queen was a Assassin Mage Defender and had skills from all those trees.

My Grinders and Dreadnaughts waded through twice their number Horsemen and militia with repeated ease. Pariden did surrender unlike Ceresa though.

By the time I finished her off though Gilden had taken out Alter and Tarth and was trying to crush Krax. Gilden, as usual, was the strongest faction by far. I decided to take him out. I built of 3-4 armies led by champions full of Dreadnaughts and some new chain-mail units called Greywards. By this time I had a some very high level fortresses that could pump out even Grey Wards and Dreadnaughts in a single turn. Kinda OP. So I just crushed Gilden using waves of units and Autocalc. The only exception was their Capital which was bugged and had something like 9 catapults in it. I took out all his other cities and forced him to surrender. Then declared war on the pitifully small faction of Krax and immediately got him to surrender.

Magnar was thus the only one left. Using the strategy I beat Gilden with worked here as well and he surrendered, leading to my victory.

Yeah me!

12,994 views 9 replies
Reply #1 Top

Conclusions and Analysis.

-Every base melee unit needs strength. At 4 points there is no reason not to give it to militia, horsemen, spearmen , or any unit that will have any armor. Just this small change will help the AI a fair bit. There should be no available melee unit designs without strength.  Even with leather armor it will boost initiative by 2. Alternatively you could get rid of it and give all units +20 WC, then add a new trait that give -20WC, and reduce the trait slots to 2. 

 

Reply #2 Top


...Even with my army right beside their town the AI sends out a group of horsemen to take back that outpost, instead of maybe attacking my solo units. They also placed their sovereign alone in the middle of my guys... with no back up what so ever. This unfortunately happens all the time. The AI just does not realize that putting small groups of units or even solo champions right next to my huge armies is a bad idea.
End of quote

Yes, I have noticed the AI do these sorts of things and was going to point them out as well.

If the AI were to better group it's units and not send small groups out close to the player's large army, I feel it could be much more effective.  Also, I have sometimes seen a good sized army that doesn't seem to have a direction and just wanders around (at least in my latest game).

Reply #3 Top

Very nice analysis, makes me want to write one. Probably wait for 0.52 though.

By this time I had a some very high level fortresses that could pump out even Grey Wards and Dreadnaughts in a single turn. Kinda OP.
End of quote

Have the same problem. I like to create mounted chainmail paladin units with 3 traits and artifacts and steamroll the world. I only ever run out of crystal and can build them in a single turn. 

Either labor prices need to go up on traits and items, or armories and barracks are too good.

The AI just does not realize that putting small groups of units or even solo champions right next to my huge armies is a bad idea. Kinda a huge deal
End of quote

This needs fixed if we are ever to take AI seriously. It needs to be able to avoid armies that it has no chance of beating and to never let heroes walk alone.

 

Reply #4 Top

Quoting Stupidity10, reply 3
Either labor prices need to go up on traits and items, or armories and barracks are too good.
End of Stupidity10's quote

I think it is a combination of the unit cost reduction improvements and the super high production in Forts, not a problem with unit costs.

Reply #5 Top

Quoting DsRaider, reply 1
Alternatively you could get rid of it and give all units +20 WC, then add a new trait that give -20WC, and reduce the trait slots to 2.
End of DsRaider's quote

I think that is a great suggestion, because it would be much easier to balance the traits with only two slots and the trait selection would be more difficult.

Reply #6 Top

I would really suggest against making far-flung conclusions without considering the entirety of the game's systems.

Not all factions are designed like Ythril.

Is Ythril powerful at yielding large, well armed armies?  Yes.  Does that mean because you played strategically and built up some nice cities with high production that they need to somehow nerf production?  Absolutely not.

Consider both Tarth and Kraxis who have unrest penalties (+10% and +5% respectively.)  Nerfing production would drastically affect both of these factions.

And I also disagree with your conclusion about giving trained units +20 WC across the board, then limiting their trait picks to two.  Your reasoning is absolutely backwards on this.  It would simply make Ythril that much more powerful. 

I would, however, like to see SD take a balance pass on the traits, and I would also suggest they allow traits to be picked more than once.  They also may want to take a look at mid to late-game unit wages as they seem a bit low right now.

Reply #7 Top

Quoting mqpiffle, reply 6
Not all factions are designed like Ythril.
End of mqpiffle's quote

I am aware. It is very easy to compensate for the difference though. I am aware of the differences and have played as every race.

Quoting mqpiffle, reply 6
Consider both Tarth and Kraxis who have unrest penalties (+10% and +5% respectively.) Nerfing production would drastically affect both of these factions.
End of mqpiffle's quote

 A 5-10% difference means nothing when you are building units in a single turn.

Quoting mqpiffle, reply 6
And I also disagree with your conclusion about giving trained units +20 WC across the board, then limiting their trait picks to two. Your reasoning is absolutely backwards on this. It would just make Yithril that much more powerful first of all, and limiting choices even further would be ridiculous.
End of mqpiffle's quote

How would it benefit Yithril more then anyone else? It would simply give every unit strength. Which is what usually happens anyway.

Reply #8 Top

Quoting mqpiffle, reply 6
Is Ythril powerful at yielding large, well armed armies? Yes.
End of mqpiffle's quote

No, no they're not.  Yithril is average at troop creation and that's the issue. It's not difficult to build forts that can pump out any troop in one turn and give them huge bonuses (Ive done it with Altair and Magnar). Almost any lv3 fort can do it. Labor costs are irrelevant to anyone with a decent fort, even the mediocre one take only 2 turns. A 10% increase in unrest is also negligible as it's so easily countered by just a couple buildings, which forts build quickly.

Either labor prices need to go up on traits and items, or armories and barracks are too good. Armories and barracks start offering like 50% troop labor cost reduction at high levels.

 

Reply #9 Top

ploughed

Quoting DsRaider, reply 1

Conclusions and Analysis.

-Every base melee unit needs strength. At 4 points there is no reason not to give it to militia, horsemen, spearmen , or any unit that will have any armor. Just this small change will help the AI a fair bit. There should be no available melee unit designs without strength.  Even with leather armor it will boost initiative by 2. Alternatively you could get rid of it and give all units +20 WC, then add a new trait that give -20WC, and reduce the trait slots to 2. 

 
End of DsRaider's quote

 

My unit designs almost never have Strength (+20 weight), although I do use Muscle (the +15 weight, +1 attack). Compared to Fast (+2 initiative), strength sometimes fails to provide bonus initiative, and fast always does. One can often carefully tweak armour to end up with 40% or just under 80% encumbrance. And you generally get more armour at 70+ encumbrance with Fast, than you do at 40 encumbrance with Strength.

That said, the units that the AI can use should be carefully designed so that they have sensible encumbrance breakpoints to make their initiative sensible.