• Foofu

...

 

Tamral

I've tanked Tidalvess with as few as 11900 HP unbuffed (about 16 - 16.5k buffed) to around 14k unbuffed (around 18.5k buffed). Never had an issue with him. The only fights I've seen in there where a deep deep health pool pays the best benefits were Vashj and KT.

It's not at all a knock against stam. If I knew I'd be a tank when I first rolled my character, I would have been a human or Tauren rather than a night elf (5 expertise or 5% more health scales phenomenally better for tanking than 1% dodge)

That being said, it's just a question of what is the best way to avoid the spike damage from occurring in the big place.

Think of it this way. For spending a few itemization points to limit the risk of being whacked really fast in succession, you don't have to overcompensate as much for the times that it does happen.

 

Veneretio

The assumption in your math that all warriors will have (or should have) the same armor greatly skews your results. I realize that incorporating armor into your calculations would make it much more complicated, but without doing so your results have an extreme bias and while still offering some interesting insight they simply fail to paint a complete picture of the strongest way to gear yourself for an encounter.

 

Tamral

It does not assume any of the such. It simply assumes a set of baseline stats, and gives you the information on which stats are the best ways to spend itemization points. Let's take the extreme end. If all 1000 or so interchangeable itemization points were spent on armor, assume the baseline stats and 29k armor. Just plug in the known numbers and it immediately reveals itself as not very viable. The end effect is baseline everywhere else in exchange for 20% less damage per incoming strike. While on the surface it seems safer, those same itemization points all spent on stamina (which is also not possible), would lead to a baseline build with around 29k HP, which ALSO registers as not optimal.

Granted if a boss existed that hit for 12k per hit or so on 17k armor, then yes, you would see stamina play a more prominant role to bring that value up to 27k or so.

but in the end, bumping up armor vs increased stam is an easily won argument. For the same itemization points that reduce incoming damage by about 20%, you can double the health pool. Since in any stream of combat that does not enter subset 3 of state 4, (including hit, hit), healing > incoming damage, the start of the next cycle is inherently in less danger with an obscene amount of health in the pool to recover to, then simple mitigation.

It is not just damage that nukes a tank. It is SPIKE damage that nukes a tank. Regulating and minimizing spike damage is the primary key. The rest is elementary, and as the OP said, just pressing 54445424452, etc..

 

◊ Xerophyte

Well, he's right to the extent that if we were allowed to pick our stats for a given item budget freely we should generally not gear for bonus armour. However, in game we can't pick from a continuous distribution of budgets, we have discrete pieces of loot to get. Yes, stamina on gloves would be better than armour on gloves if both enchants had the same item budget, however as the armour one has a larger budget we should still pick it over stamina for any physical damage encounter. Yes, armour on rings isn't a strong stat, however [Violet Signet of the Great Protector] is still not replaceable for a low-threat, high-mitigation/avoidance scenario until T6 content.

It's a very good piece of analysis and the assumptions are mostly sound, with the possible aside of healing which is going to be a great deal more stochastic than is modeled. How applicable the results are for actual in-game gear selection is something I really question, though. Yes, it's good to know that the perfect helm of a given item budget has spent said budget on a balance of stamina & avoidance stats. However, the decisions I'll actually have to make in game at the T6 level is between 5 (or so) different helms ranging between ilvl 127 (Tankatronic) to 151 (Impenetrable) and budget use ranging from excellent (Tankatronic), through good (T5) to confused and inefficient (T6). Actual gear selection requires having decent stat weights even for stats that are highly inefficient to budget compared to other stats.

 

Sepulture

Ok, I swore to myself that I wouldn't jump into this one because I want to stay employed, but I can't resist.

It's a nice model, I think it can be used as an indicator for gearing, but there are a few flaws (as with all things modeled).

1) Healing assumptions
Model- assumes steady HPS for healing
Reality- Healing is proactive based on known abilities that occur at regular intervals (RoF, Cleave, Arcing Smash, etc.). This results in healing spikes around known damage spikes. Healing is also reactive based on damage spikes. The reactive lag is = latency + reaction time + cast time.

The point here is, you can't assume a steady HPS as yin to your unmitigated/unavoided damage yang. Healing is not a big HoT. It's quite spikey. Assuming a HoT model for healing is a huge avoidance bias, because large healing spikes don't do as much for a tank with a smaller health pool.

2) Damage assumptions
Model- assumes autoattack. Physical
Reality- in addition to autoattack, bosses use both timed and "AI timed" abilities which have special characteristics and aren't always physical in nature.

The point here was illuminated in the discussions about that shady Karathress shaman and Kael'thas. Spell damage changes everything, and when a big part of your spike is spell related, or even physical occurring at a known interval, that favors stamina, or in the latter case stamina/mitigation.

I know it's much easier to find flaws in something than it is to create something, and I think that model is surely a good indicator. Still, there is a lot to be said for combat log parsing and knowing your gear, healers, and adversary.

 

◊ Quigon

Yeah the violet signet is very nice - and armor will absolutely reduce damage - avoidance and stamina cannot say the same thing. On a lot of calculators armor ends up being one of the weakest items, but the fact is without it you'd be a mage with avoidance. Armor is underrated, and very valuable - but it just turns out that in tier 6 gear you can get absurd and almost broken levels of avoidance right now (incidentally this usually happens in the highest armor gear too).

Also, I have seen no evidence that the FLK mobs share aggro. They would be the first mobs of their kind outside of thaddius's adds, which come with a specific script event when this happens.
You can pickup aggro over healers with very little effort on nearly any boss when the tank dies - First, I thought healing aggro was divided out, and second, we've done this on kael'thas when the MT died at least twice. Outpacing even a long period of heal aggro on some bosses is readily done. That does not mean that aggro is shared, in fact, I think you can readily prove it is not shared because the bosses don't attack the offtanks who unload on the boss during DPS phase.
Besides how would aggro even be shared? The game is distinguishing tanks? Warriors? Ferals? Its giving all 25 raid members the same aggro? Its just not realistic.
If aggro was shared on the mobs, your DPS would be dead - it would go to the mage or rogue doing a ton of aggro on each boss, moreso total than any individual tank, but lesso on each individual boss. Aggro rules are typically explained without radically changing sharing rules (like 4H).

 

◊ Quigon

I updated the latex to mathtext that these forums are using now. For anyone interested in those equations they should be working now. Plus added a few major updates based on user feedback.

Would anyone here be interested in writing the a "DPS Warrior" tanking section?

Simply the skills and abilities one would use as a DPS warrior in an appropriate tanking rotation - with some explanation to decision making, ideal situations, and perhaps variation based on spec and situation.
If no one bites I'll ultimately do it, but people with more experience tanking as these talent specs would probably do a better job in a practical sense.

If you're going to put a lot of effort into this, post here before you write it, so that multiple people don't waste time doing the same thing.

I'm still going to add the "How to DPS as a prot warrior" part soon.
Edgewalker has done over 1000 DPS as prot, and we're looking to get either a movie or a good writeup from there.

 

galzohar

Assuming a healer is doing something low like 1000 HPS, with salv that would be 300 TPS. Say it's shared on 4 mobs that's 75 TPS. Over 60 seconds it's just 4500 threat which is not impossible to overcome, but consider longer duration, more threat when one of the adds is dead, and most likely higher HPS and no salv on the healers and it becomes that much harder to pull it off before a couple of healers are dead. Not to mention your TPS won't exactly be 1000+ without a boss hitting you for rage. I'd like to see a more detailed explanation of how you get aggro when the MT died on kael and karathress (assuming they're taunt immune, of course).


Linear scaling stats and their relative increase
A couple of pages ago someone showed diminishing returns on armor. However, this can be said about *ANY* stat that scales linearly. Call the stat X and benefit=a*X+b where b and a are constants, the relative incerase of adding dx would be dx/(ax+b) which obviously goes down as x increases. So saying armor has diminishing returns is kinda misleading, as every stat in the game that scales linearly (and most stats in the game scale (at least approximately) linearly). That kind of diminishing return can be said about increasing stamina for survivability, block value for threat or attack power for DPS. They all have diminishing returns on their relative increase by increasing the stat by dx, since they all scale linearly (or approximately linearly).

 

◊ Quigon

This is a little bit disconcerting. I would say fairly confidently: no on both accounts. First of all, your derivative would lead to a scalar, second of all it is done in a vacuum and does not make practical sense.

Block value is a scalar increase - yet it provides exponential returns per point. This is straightforward:

If a boss hits you for 1000 and you block for 600, you take 40% damage. Increase 200 block:
If a boss hits you for 1000 and you block for 800, you take 20% damage, a 100% improvement from before, for only 33% more block value. Increase 200 block:
If a boss hits you for 1000 and you block for 1000, you take 0% damage, an infinite increase, for 25% more block value.

Mitigation and avoidance has exponential returns per point for their relative damage reduction. If you go from 99% to 100% mitigation or avoidance, you become invincible.

The problem is mitigation takes exponential amounts of armor to increase per point. This is unlike any other stat that I am aware of, unless you are defining "slow" abilities by absolute reduction - a change sometime around the Naxx era.

The math done on the original post, and in the one you described show the precise relationship between the change in relative damage reduction as a function of armor, instead of mitigation which is non linear.
Armor vs mitigation defeats mitigation vs relative increase by a square factor.

Avoidance returns are exponential, period. There are no diminishing returns.
This is also shown on the original graph.

Also FLK mobs do not share aggro. They just don't.
Why would the mobs go to the TANK if they shared aggro? The DPS do more aggro overall to the sub bosses than any individual tank, even if each individual tank themselves does the most on a particular mob.
The thing is is that it is trivial to overcome healing aggro on a boss even after the tank dies - it always has been that way!

We've had the FLK tank die when the THIRD add went down, about 4 minutes into the fight.
The boss then went straight for the offtank healer healing the third sub-boss (this tank was primarily healed by 1 player, whereas the MT was healed by 4-5 players so the mob went to the healer who had the highest load during that entire time).
It killed this healer, then another, and by then I had landed 2-3 shield slams, and a few other moves, and he stuck to me for the rest of the fight.
I'd say that is pretty strong evidence that aggro is not shared, because it went to a healer first, not a single dps, not a single tank, except the tank who was attacking him, and only after killing the top aggro healers.
Classic aggro situation.

 

Yaddle

XVI: Raid Specific Buffs

I only dabble on my druid alt who is resto with a touch of balance. Is Insect Swarm not worth mentioning?

 

◊ Ulfgar

Karathress certainly isn't tauntable.

Key in getting another tank top of the aggro list on a tank death in that situation, at least in my experience, is chain Misdirection.

 

Tamral

Well, also keep in mind that the itemization is based on an initial value in armor of 17000. If you started everything at 0, armor and stam would stack to pretty solid levels before a single additional stat was even considered. As for the concerns regarding healing building itself around spike and special abilities, you are absolutely correct. Many times I have seen tanks drop it was not to any of the special ability spikes. It was usually to a parry induced spike. You find me a healer good enough to recognize that one coming and prepare for it, and I'd be extremely impressed.

 

• Branar

Yeah, I've never seen any evidence of that either. I think it's such an easy encounter to beat healers on agro because the healing agro is split 5 ways. In fact, it may be split more than 5 ways, since I suspect it's likely that totems count as mobs, and the whirlwind thing that the priest add summons probably counts as a mob as well. So you have FLK, his 3 adds, the hunter's pet, 1-3 shaman totems for most of the healing done, and that tornado.

So your healers are getting anywhere from 1/6 to 1/9 of their normal agro on every heal on each mob, depending on how many totems and whatnot are down. That's not a whole lot of agro and it's easy to see why tanks can overcome it, given how little agro healing is to begin with.

 

• Polleke

Sorry to derail this even more in a threat discussion. But global tanking threat would then also be split right? It doesn't explain why if a tank dies the mob would come running to another tank. The global threat (shouts) that this warrior would have generated should have been split just as much as the healers.

Back to ontopic. This thread gave me the motivation to level my warrior. Thanks for the excellent information.

* Bla

 

◊ Xerophyte

Bear in mind that any healing done by the tank - earth shield, lifebloom blooms, prayer of mending, judgment of light, leader of the pack, etc - is subject to their threat mod of 1.5. A healer with bos & a -threat talent will have a threat mod of 0.56. Now, I don't know how you folks end up but our tanks will typically do about a quarter of the measured healing done by an actual healer with various self-heals. This leaves a relatively small threat difference to begin with and it's certainly plausible that on some fights you'll have done more than average self-healing which along with threat from rage gain effects (shield block, bloodrage), buff threat from shouts and so on makes you actually outpace healing threat. It wont be very common but having it happen every now and then isn't really shocking.

 

• Xaviera

DPS'ing as a protection warrior always seemed to be something pretty simple. I've done it, and I've done well over 1000 dps as well, as I'm sure anyone has with decent enough gear.

Not too different than gearing for a DW Fury set, except as Prot spec stacking items with specifically strength is really beneficial because of the extra 10% we get from that.

Hit rating is pointless past getting to the hit cap, you're much better off spending any itemization points on pure Strength at that point. I also geared with the assumption that Crit and Haste aren't as good as strength in terms of itemization points spent on. A prot warrior gets no 'extra' benefit from a crit, no deep wounds, no impale bonus, no flurry, just double damage, which is going to be better than extra hit, at least.

Really slow mainhand, get the hardest hitting weapon you can find. My offhand is also the same thing (Dual [Blade of Infamy]). I have the mainhand enchanted with Crusader, and the offhand with Potency. Crusader has a 100% uptime nearly and the strength/AP gain is massive. The offhand is hitting so unfrequently that nothing with a proc will be up nearly enough to outweigh the gain from potency.

Having two slow weapons as your setup increases the damage from Whirlwind a bit, one of the few abilities in your 'rotation'. And since you don't have flurry or anything that benefits from rate of attacks, I see no specific gain in using a fast offhand, unless it's just an extremely superior weapon in every way to a slow.

Rotation-wise Devastate on every cooldown, same with Whirlwind, and continue to use Heroic as a rage dump.
Don't neglect Bloodrage or Berserker Rage, assuming the fight you're on has some amount of raidwide damage. Gorefiend is probably the most popular fight for an extra prot warrior to be dps'ing on.

If there's something glaring I'm missing then I suppose it would be nice to get more input on it, but it always seemed rather straightforward to me

 

galzohar

Which is exactly why armor is a linear stat, just like stamina, strength, attack power, spell dmg... Of course for some classes in certain cases the increase isn't completely linear, but for most stats in the game for most cases it's pretty close to linear. Armor is no different, as the amount of unmitigated damage you need to take before you are killed increases linearly with armor.

Of course avoidance has exponential returns when it comes to "with X points of healing available, how long would it take me to die?" however it does nothing when the worst case scenario comes up, and it will come up sooner or later. The only way I could see avoidance being a good thing is if a burst that can kill you is so rare that it will happen on a low % of the full fights (so, say, you define it as "OK" to get gibbed once every 10, 100 or whatever fights), and the biggest burst that is likely enough to happen is too small to be a concern.


The reason for every linear stat having its relative increase with diminishing returns is easy to explain without math really - when you have 50k HP 5 more HP isn't as noticeable of a benefit as if you had 10 HP. Relative increase is increase/current, which for a fixed increase goes down as your current goes up.
Another way to look at it is that armor multiplies your HP in terms of "how much unmitigated damage you need to take to die", so same amount of armor is worth more with more stamina. This works the other way around, stamina being better when you have more armor.

As for BV, I it's linear for threat, not survivability. For survivability it's exponential, yes, but not as much as you think. In fact with most bosses the survivability gain from block value is quite close to linear.
Just like saying that 10%->11% avoidance will give 1% less damage taken is not far off at all. Same with BV, since you can almost say that BV<<dmg then you can say the increase to survivability from BV is not very far from being linear. If you tank something that hits for something closer to how much you can block then yeah it's going to be a lot more far off.

 

• Xaviera

Avoidance is a good thing because it reduces damage taken. Armor is a good thing because it reduces damage taken. There's not a lot of stuff you can do to stack armor though outside of wearing your highest ilvl plate with armor enchants, rings/cloak/weapon, etc. Whereas there's a lot of choices in the rest of the slots to use Stamina or Avoidance. Choosing avoidance in those spots will cause you to take less damage, so that's when it's smart to do so after everyone's comfortable with the fight.

If there was a lot more itemization possibilities for armor (Armor gems, armor necks, extra tank gear pieces with itemization points spent on bonus armor, etc), it would be great, but because there isn't, once you're trying to minimize total damage taken, you're going to have to use avoidance pieces.

 

◊ Quigon

Galzohar:
Like a previous poster: Don't knock it till you try it.
It turns out the best avoidance gear usually has the best armor as well.
And nearly the best stamina. You can stack avoidance to almost exploitive levels. It goes far beyond being academic.

Stamina was great for a while during this game, but right now avoidance is really powerful. Gotta learn to adapt and try new things. Trust me, avoidance is amazing right now, at least for some fights.

Also this linear argument is getting a bit semantic and saying something is kinda linear and not exponential does not make it linear, although I do agree with your point on Block value, but my point was made to counter argue your original post, and taken to a reasonable extreme. And armor for instance is not linear either, this one is definitely noticeable.
Stamina I agree with to an extent - the equation for that is given on page 9 of this discussion I believe.

Your comments about the mob getting lucky are well discussed in the guide already. This was afterall part of the original discussion on the conquest thread. Remember getting avoidance back then though usually meant wearing greens "of defense" or something (although that was nerfed).

 

galzohar

Can't see your armory profile but my guess is that you're already farming all the game content. This means you probably already have more than enough stamina/armor/BV to survive anything within reason (as in that doesn't have only 0.1% chance to happen per fight) which makes avoidance feel a lot more confortable. I'm just not sure a progressing guild will get the same kind of experience, when the gear level is lower and having enough burst survivability is not a given.
I'm not ruling it out completely, as i've yet to be in hyjal/BT, but any encounter I've done so far have seemed like the tank could've used more stamina. Even when you go through a "omg free badges" heroic mech with no CC if the tank has only 12-14k HP he can get gibbed faster than you can land a heal with some double-pyro+2 mobs melee. 15k HP and you suddenly have so much more breathing room, which is less than what switching to 2Xdarkmoon would provide. And we all heard about tanks getting gibbed by windfury on karathress.

Then again I can see your point that above a certain point of burst survivability where more stamina is really going to become useful only on very rare occasions (all healers had to move at the same time for too long, 2-3Xparry+instant or something like that) you could actually start arguing for avoidance to make healer's life easier. Determining what that point is is harder to do, but if I had to guess, on a boss with no specials have at least enough to survive block+crush when your heatlh deficit is the size of a FoL would be something to shoot for? Meaning 6k normal+9k crush with the armor you have and paladins FoLing for 1.8k would mean 16.8k HP on that boss would be a bare minimum, less than that your avoidance:stamina tradeoff is starting to be a lot less worthwhile. Any thoughts about this? How would you determine the point where "here I need stamina" and "there I have enough stamina - avoidance would be better"? as I think we all agree that the 98% avoidance tank would not be viable even if he could do insane therat just becuase he can get gibbed too easily, and while it may take it a while to happen, it's not unlikely to happen on a lot of the fights, at least those with magic damage involved.