There are so many threads discussing what is and what isn't powerful in Skyrim that I decided to play Mythbuster and actually test it. To do so, I created a mod called Battle Balance Arena (links at the bottom) that pits NPCs of various archetypes against one another and used it to simulate a ton of battles.

Before I go into discussing the way the data was gathered, I would like you to acquaint yourself with some big red disclaimers:

DISCLAIMER 1: What I tested was battle balance, which I define as the ability to deal and withstand damage in a fair, direct fight. Within this definition shouts, stealth, summoning, illusion, healing and consumables are all ways to avoid being good at combat. That's why they were excluded from the tests.
DISCLAIMER 2: The exact method that I used is just one of a million. If you want to test balance in another way, feel free to use and modify my mod for that purpose.
DISCLAIMER 3: It is perfectly reasonable if you run the exact same tests yourself and get slightly different results. I just don't have the will or time to gather a sample size large enough that things like AI decisions and dice rolls on perks are completely accounted for.

Now that we got that out of the way, here is a rundown of the way the testing was performed:

6 different teams, each of which represents a combat style, fought against each other in every possible combination of two (a total of 15 matchups) as well as in a special "Nemesis" format against a boss monster. Each of the matchups was done in two formats - solo and group, across three ranks - Rookie, Veteran and Master. After every such iteration one or both of the teams scored points. A team's overall result is the total score across all formats, ranks and matchups.

The teams:

• Knights (KN) - heavily armored fighters that use a shield and mace.
  
• Barbarians (BR) - heavily armored fighters that use a two-handed battleaxe.
  
• Berserkers (BS) - heavily armored dual-wielding swordfighters.
  
• Archers (AR) - lightly armored archers. They also have a dagger for when forced into melee.
  
• Mages (MG) - stereotypical Skyrim mages that use Frost spells, magic armor and wards.
  
• Spellswords (SS) - heavily armored fighters that use a sword in one hand and Fire spell in the other.

Each of these have perks and stats the same as a Player Character would, as well as enchanted equipment taken from leveled lists.

The ranks:

• Rookie - level 1, no perks, all stats at 100, starting equipment (iron and hide), no enchantments. Imitates characters at the start of the game.
  
• Veteran - level 26, 25 perks, 100 points spent on Health and the other 150 according to archetype, skills at level 60, enchanted Orcish and Scaled armor, no ring or necklace. Imitates the characters as they would be halfway towards the soft cap.
  
• Master - level 51, 50 perks, 200 points spent on Health and the other 300 according to archetype, skills at level 100, ennchanted Daedric and Dragonscale armor, enchanted ring and necklace. Imitates end-game characters.

When choosing enchantments, a rule was followed not to stack the same enchantment, and only use Resist Magic (elemental resistances not allowed) for magic protection.

The formats:

• Solo - begins with a best of 6 in a 1v1. If the result is a tie, both teams get 1 point. If not, a second round is played, with the team that lost the first round bolstered by 1 extra contestant. If the result is a tie or in favor of that team, the match ends and the team that won the first round gets points equal to the number of contestants on the opposing team. If not, the losing team gets another extra member and another best of 6 is played, and so on. The maximum is 10 contestants. If even then the losing team is beaten, the winner gets 11 points.
  
• Group - begins with a best of 4 in a 4v4. If the result is a tie, both team get 1 point. If not, a second round is played, with the team that lost the first round bolstered by 1 extra contestant. If the result is a tie or in favor of that team, the match ends and the team that won the first round gets points equal to the number of contestants on the opposing team minus 4. If not, the losing team gets another extra member and another best of 4 is played, and so on. The maximum is 10 contestants. If even then the losing team is beaten, the winner gets 7 points.
  
• Nemesis - in this special format the teams fight not against each other, but against a Draugr boss. It starts out in a 1v1 and the team wins a round if it wins 3 times out of 3, If not, it is bolstered by 1 contestant and another round is played. Once all teams have fought the same boss on the same rank, the team that took the most number of fighters to beat the boss gets no points, while all others get points equal to difference between that and how many members it took them, multiplied by 3.

That sums up the method. If you want specific details, open up the mod in the Creation Kit to see what went where.

As for the results, they paint a very clear picture:

https://dl.dropbox.com/u/36077861/BBA%20test%20results.jpg

My observations:

• The best combat styles are Spellsword and Berserker.
  
• By far the worst is the Archer, and the pure Mage doesn't fare much better.
  
• Barbarians and Mages get worse as they level, while Knights and Berserkers get better.
  
• Mages benefit most from fighting in a group, while Berserkers are the best solo fighters.

My conclusions:

1. Melee damage is overpowered.
This is proven by both the fact that characters that don't block have an advantage over characters that do and that the Spellsword and Mage are worlds apart. It seems that maximum aggro up-front is the formula for success in Skyrim.

2. The community is right to consider mages greatly underpowered.
Not only are they near the bottom of the ladder, they also get weaker as they level up. This is abated somewhat if they fight in a group, but it still isn't enough to keep pace with the other archetypes.

3. Contrary to popular belief, the worst archetype for direct battle isn't the Mage, but the Archer.
Archers just cannot fight directly without having some kind of advantage, be it crazy distance, stealth or poisons.


Links to the mod:
Steam Workshop: http://steamcommunity.com/sharedfiles/filedetails/?id=76818896
Skyrim Nexus: http://skyrim.nexusmods.com/mods/18782/

Discuss away.

Sounds nice.. But I've heard that Destruction scales with NPC's while they level but not with the player... mayby something to think about..

thats the point on why it is very bad.

The archer was the most underpowered? Uh oh, I just started a stealth archer character. Hopefully I have better luck

It does seem like quite extensive research, although the only light armored team in there was Archers...
It does pretty much confirm what was mostly suspected among the players, did each team only have 1 style of weapon or spell each.. or was there for example Knights with Axes, Sword and Maces?


I believe the test was for direct combat only, so as a stealth player, you can kill the enemy before he gets to you.. Also it seems if you pair your Archery with Dual Wielding you should be fine in both Stealth & direct combat

Nice tests


The tests were made in direct combat so it's normal an archer/stealthy character wouldn't go to far

Well there's one problem with these observations:
You used NPCs to fight each other, they don't tend to use the same tactics, or even the same tools as the PCs could. Especially with magic as it works differently for NPCs.

The trick is not to be seen

Seem like a cool mod, I do think there should be more classes. It doesn't go too mutch in depth like Bound Archers and such. Would have been cool if you had added unarmed.
Example-
Level 1 Unarmed- Heavy Armory -Iron No Special Abbilities
Leve 26 Unarmed- Obtained Gloves Of Puligst Orish Armor Heavy Armor that allows stronger unarmer attacks.
Level 50- Enxhanted Armor for Unarmed Attacks Deadric Rings and Necklace

Some objections:

1) As noted upthread, NPCs don't fight like or even have full access to the abilities that players do. Several abilities NPCs have will flat-out not function on a player, and NPC health scales very differently from player health. Even if we accepted your premise that an arena is an effective way to measure balance in this game, an arena of NPCs amongst themselves is certainly not it. You've measured the combat effectiveness of NPCs against each other.
2) The notion that a giant melee in a closed arena significantly reflects how combat gameplay in Skyrim works out is not something I can really swallow.

That's not to say that this isn't interesting, or informative, because it is; but the type of information obtained from this is extremely conditional and limited.

Beth should make magica regenerate at the same rate stamina does.

1) I agree that using NPCs doesn't yield the best results (hell, that's why I had to referee the matches for when the AI decided to go trololo). But that was the only way to collect any sizeable amount of data. Besides, the point was to illustrate the differences of damage output and mitigation, which holds true for when a player character plays the same combat style. Also, as far as stats and perks are concerned, I've made certain that they matched what a player character would have. And the AI's did in fact take advantage of most of them - knights used their Shield Charge, mages utilized the slow and Deep Freeze of their Frost spells and so on.
2) Please read Exclaimer no. 1 again. I was testing efficacy in direct combat. If an archetype needs particular landscape or circumstances to function, it may be a good archetype, but not one that excels in direct combat. And let me tell you, if you run the mod and pit Berserkers against anyone, you'll witness the true meaning of massacre. The damage they deal is so ridiculous most matches end in a heartbeat.

The melee classes were each equipeed with exactly 1 type of weapon. Spellcasters had several spells:
Rookie Mages had Frostbite
Veteran Mages had Ice Spike and Ice Storm
Master Mages had Icy Spear and Wall of Ice
Rookie Spellswords had Flames
Veteran Spellswords had Firebolt and Fireball
Master Spellswords had Incinerate and Fireball

Much as I'd love to test more builds and combat styles, the problem is that with each extra team the number of tests grows exponentially. If I had to do more than the 150+ battles that I already did, I would have gone bonkers . That said, feel free to modify the mod in any way you want to test what you want.

Nothing to worry about unless you make a non-stealth archer.

While a straight up archer may be "weak." A stealth archer is borderline for being the most powerful thing in the game given that it gets to a point where you're invisible even when some enemies ought to be able to see you once your sneak is high enough.

Of course a rouge type dosn't win against warriors straight up.

That's not what being a rouge is all about.

The problem with this is that this is based on 'run in all guns blazing and hope for the best'. That might be fine for a warrior, but a mage, archer, thief, assassin and other archetypes will require more tactics, as well as a mixture of different skills.

Therefore, I wouldn't call this definitive by any means. Although well done for taking the time to do this

Well, that's just the thing. If archetype A needs help from a specific approach and support skills and archetype B doesn't, then people perceive B as being stronger than A. That's what a lot of the balance discussions are based on.

For the same reasons put forward by Solitudian, interesting but, rather pointless.

As compelling a read as that was... I believe you meant 'Disclaimer' and not 'Exclaimer'.

Poor old Archers though, we knew they were squishy, s'why they have to kepp out of sight/reach.

Characters that dont block have advantage over characters that block without a shield is more like it.

With the final shield perk, you do not block, you run them down like tissue paper. I "literally" "run-over" 99% of things and if it were outdoor stampede them to death on Master difficulty... tis boring... really.

Fixed.

Master Knights did in fact use Shield Charge, but only occasionally. That's one of the benefits of using AI's - they don't abuse exploits.

I always said spellsword was one of the more effective combat choices, even though others said otherwise, or atleast used to, I have been away for a few months. Though with heavy armor I consider that more of a battlemage, and I see spellswords as light armored sword and spell warriors.

Though as those before have said, this is interesting but doesn't have much direct application as far as a player character vs the NPC world of skyrim.

I really don't think mages are as weak as they always portrayed, yes it doesn't have the raw power of a dual sword user or 2h warrior,

but how would have things played out if you used conjuration on the mage, as so many NPC mages do? I do agree with the use of frost magic as I believe that will ultimately lower damage taken, even if it is slight, in the long run. Also, How would it have fared with fire magic?

As underpowered as destruction is, I have always been able to do well with it, even on master, in fact, up to level 20 or so, I think a destruction mage is easier to succeed with over a melee warrior. and really, magic only is underpowered compared to melee when coupled with smithing, which i usually avoid until later levels, and use alchemy and enchanting without combining the two, nor will i combine then with smithing anymore. Destruction magic paired with alchemy is pretty insane by the way and id say its far from underpowered.

heh, i haven't played in a few months and the character im playing now is a level 32 paladin type character, 1h and restoration, heavy armor and alteration for defense, all this talk of magic is making me want to make another mage or spellsword!

Yes... you are right. The Shield Charge can be abused madly by players. Guilty as charged =P

I find that from Daggerfall to Skyrim (havent really played Arena to judge), Melee is usually overpowering in the end and I find that I have always been sort of a BattleMage - Heavy armored sword swinging spell caster.

I can understand why it is so though, because mages are more common than Wolves and Trolls in TES; especially Skyrim, they cannot be too strong. Bethesda like to throw half a dozen of mages at players at many point, if they are powerful, it is gonna be a nightmare >.<"

Sorry if I came off a bit rude there BTW. I feel a bit bad picking up on that after you went to all that trouble

Funnily enough... as the PC I find that mid-late game, high level archers do the most damage to me when compared to any other NPC type... If I'm not quick enough to dodge their arrows that is!

Nice tests! Definitely gives some perspective over the effectiveness of NPCs using melee weapons...

An improvement to this would be to use the mod to test the different types of melee weapons (with their perks) exclusively. Maybe with light/heavy armour variants too...
Unfortuantely I'm an Xbox user and can't apply such mods...

I'd envision my test like this:
Teams should be identical with one change only (as in all true science experiments! );

I'd pit 9 teams against each other. 2 vs 2 would likely be best (1 member light armour, one member heavy armour). Each team of that type would have a different weapon type and the appropriate perks (axe/sword/hammer) Dual perks would be added to dual wielders, and the equivelant number of block "bashing" perks would be added to the others:
3x Teams of One-handed shield wielders
3x Teams of One-handed dual wielders
3x Teams of Two-handed

i.e. Team 1 would be 2x One handed Swordsmen with shields, with the first one-handed perk (+20%), the 3x blade perks, and 3x shield perks (bashing ones). One member in light armour, one in heavy armour. (posisbly a couple of perks in those too!)

That would be my runthrough.

-HK

I found that all melee AI's use the same strat against archers - they sprint to close the gap ASAP, usually tanking 1 shot in the process. This forces the archers to sheathe their bows, at which point they've essentially lost. To be effective, they need cannon fodder to direct attention away from them, which they typically have in an average encounter in Skyrim. Whenever they don't, they're screwed.

I may do this sometime in the future as that would be an interesting comparison. The general consensus is that blades are the best for the late-game because their damage scales well. We'll just see about that .

I was thinking of coating spellswords in light armor and giving them magic armor spells, but I concluded that there is no real difference between that and just using heavy armor with no extra armor rating.

You weren't rude. It's just that it was late when I wrote this thread and my head wasn't performing well .