Why We Do What We Do (An in depth explanation of EHP and ICC 3.3.3 tanking mechanics)

[Hypathia]

That's what my current results tends to show (it's still need to be completed and verified) if you have full mana healers and some other conditions. Gem of Stamina > gem of dodge from simulation.
Now, there are two ways to approach the relative value of stamina versus dodge.

• Probabilistic, unfortunately this is very hard from my point of view to interpret correctly although it's really easy to have an opinion and go in the wrong direction. I would say that qualitative interpretation before simulation craft (of course there were formulation spreadsheets but it's way harder to use here) was mostly empirical.
• Simulation, where we don't really understand where the results come from. We hit on the run button and have something, a value which is like an UFO ! Only full trust in a program and small interpretation is involved.

For WotLK, the good choice was to gem stamina, although I think that in some situations, you could go for stamina/dodge for gemming bonus.


What should be nice for Cataclysm would be to develop both approaches, and be able to explain in a probabilistic way the simulations several people are doing. I hope that the way to find the relative ratio will reserves us some surprise, with let's say real choices to be made.

[Thaurissan]

I think I stated it before and maybe it was missed, but I will state this again:

My biggest question is not about stamina vs. dodge. It's about ARMOR vs. dodge. Since armor and dodge both protect only against physical damage, I think it is a much closer call than it's being made out to be. Lumping armor in with stamina as "EHP" is misleading in that sense.

As was suggested above, armor gives you a 100% chance to reduce physical damage by X%.

Dodge gives you an X% chance to reduce physical damage by 100%.

Given that the statistical value of both stats is equal over a long enough period of time, what is the actual value of the "always on" attribute of armor? How much larger does a dodge bonus need to be to justify choosing a piece of gear with dodge over one with armor?

If I have a piece of gear that has 10 armor and another that increases dodge by 20%, I'd obviously be a fool not to choose the one with dodge. It's probably impossible to determine where that dividing line is in exact terms, but that's what I'm talking about.

[Kazeyonoma]

THAT is where you're making the mistake. Statistically over a long enough period of time is NOT how boss encounters work in wow. A small chain of hits in a row will kill you if you don't reduce the incoming damage enough to survive it with heals. Dodge while over a longer period of time will in fact prove on average to reduce more overall damage than Armor, does NOT save you from worst case scenarios, and NO having AD save you the first time it happens doesn't mean it won't happen again while it is on cooldown. YES you use your cd's regardless of which class you are to save yourself, but worse case scenarios happen because there are no cd's left, you're healers are moving, you're down 1 healer, you just got smacked by an undodgeable/unmitigatable dragon breath, and are about ot take another 2 hits on you, you either survive the hit guaranteed because you've got enough armor to reduce the damage to a survivable amount, or you pray that you're extra 2% dodge kicks in and you dodge that next attack otherwise you're dead meat.

ON AVERAGE, dodge is > armor. Over large/long statistical evaluations. 3-8 minute fights are NOT large statistical evaluations. Fluctuations occur CONSTANTLY in small windows like that. sure you can tell your raid leader and raiding core, that on average, you'll probably survive more often than other tanks, but every raid night, expect me to die just randomly because my statistics took a +10% margin of error. See how that hangs with them. See how that hangs with being at 10% from a progression boss kill and you plopping dead. THEN come back and explain to them that "don't worry guys, next time, i should statistically dodge 10% more times to make up for this time, so we'll be better off!"


You can't compare items from large margins of stat allocations with each other. That's silly, that'd be like comparing a 6 stam gem from TBC to a 20 dodge gem. Of course, the 20 dodge gem is better, but if you can buy 20 dodge gems, why aren't you buying 30 stam gems? You're making the same mistake here about the CTC. If you're talking about using the CTC, you're also comparing it to like level trinkets, such as the armor trinket you can get from putricide, the armor trinket you can get from RS, or hell, even the emblem armor trinket from TOGC. Obviously there's also stam trinkets to think about, but you aren't arguing about them, in which case, if you have 2 stam trinkets you should probably be using them ANYWAYS, so CTC shouldn't even be in the discussion. So based on those relatively equal value (give or take 30 ilvls) the armor ones, still win out BECAUSE of the very nature of short bursty boss fights that DON'T simulate realistic statistical data points across a long period of time.

[Kojiyama]

The other issue with that evaluation is that there is not a significant deviation in Armor, even with all the +Armor gear floating around. Armor is an awesome stat--mostly since it increases the value of everything, including avoidance--but players are already close to the 75% cap in many situations.

You also cannot gem Armor, which limits the reason for any kind of comparisons.

[Hypathia]

@Thaurissan
I'll try to answer as it sounds interesting to me.
Let's take a base of 40% avoidance and 70% for armor. What we are looking for is the death probability, not the number of damages taken.
If we look at the famous 5 consecutive hits probability in 100 hits run which is around 0,98 for 40% avoidance and 0,985 for 41% avoidance. If we look at a 4 consecutive hits in 50 runs, the difference is 0,972 against 0,977. What we find is that a 1% reduction in avoidance is not equivalent to a 1% reduction in death. It's somewhat a "probabilistic thing"

For armor, the only way we have to understand it better is converting it in stamina. Sure it depends on magical damages. In fact there's no intuitive way to show that 1% reduction of armor or stamina is equivalent in 1% reduction in death. But we know that 1% armor >> 1% stamina for physical damage.

[Thaurissan]

and NO having AD save you the first time it happens doesn't mean it won't happen again while it is on cooldown.

There are two components to Ardent Defender. One is that a hit which would otherwise kill you heals you up to 30% of your maximum health, on a 2 minute cooldown. The other is that all damage that takes you below 35% is reduced by 20%. In other words, it is as if you had an extra 20% armor and/or magic resistance whenever you go below 35% health. That effect is always on.

[Kazeyonoma]

yes, so you get bonus EH during those hits. A hit that would otherwise STILL kill you if you don't have enough survivable armor/hp is still leaving it up to chance if you go the avoidance route. I fail to see how having an automatic shield wall when low on health changes the argument I made above about averages vs. realistic results.

I really don't want to get into specific numbers just to prove that attacks can and will kill you even through AD, and relying on avoidance to save you is bad.

[Aggathon]

I agree 100% with Kaze here.

It really makes me want to fix my spreadsheet because then you could mess around with numbers and look at the minTTD/minTTDwHeals and averageTTD stats and see how they all interact.

@AD: 20% more reduced damage is not 20% less damage taken. Remeber damage reduction effects are not additive they are multiplicative.

Lets say you have ~80% reduce physical damage (raid buffed this is a decently safe assumption). So lets say a raw damage hit is 100k damage, that is if you had zero mitigation that's how hard it would hit. At 80% reduction that means you take 20k dmg. Now say you add 20% reduction to that, that's actually (1-.8)*(1-.2) = .16 meaning you'd have 84% reduction, only 4 more percent and that 100k hit means you take 16k damage instead of 20k, only a 4k damage difference, not a 20k difference as if it were additive.

While yes, 4k less damage is significant, if you're taking a worse case scenario you're gonna want to make that extra 4k hit points go as far as possible, and if you're taking another huge hit, you'll probably die anyways. AD is not immortality, but another effective cooldown and the higher your stam and damage reduction is the BETTER that gets. It procs when you need it, dodge doesn't always happen when you need it.

I mean... I outline all of this, I'd highly suggest re-reading section 1 and section 6 more in depth.

If you have any more questions after that I welcome you to send me a PM and I can explain it further, especially since I now have experience from both sides of the coin, the healing side and the tanking side.

[Thaurissan]

I agree 100% with Kaze here.

It really makes me want to fix my spreadsheet because then you could mess around with numbers and look at the minTTD/minTTDwHeals and averageTTD stats and see how they all interact.

@AD: 20% more reduced damage is not 20% less damage taken. Remeber damage reduction effects are not additive they are multiplicative.

Lets say you have ~80% reduce physical damage (raid buffed this is a decently safe assumption). So lets say a raw damage hit is 100k damage, that is if you had zero mitigation that's how hard it would hit. At 80% reduction that means you take 20k dmg. Now say you add 20% reduction to that, that's actually (1-.8)*(1-.2) = .16 meaning you'd have 84% reduction, only 4 more percent and that 100k hit means you take 16k damage instead of 20k, only a 4k damage difference, not a 20k difference as if it were additive.

While yes, 4k less damage is significant, if you're taking a worse case scenario you're gonna want to make that extra 4k hit points go as far as possible, and if you're taking another huge hit, you'll probably die anyways. AD is not immortality, but another effective cooldown and the higher your stam and damage reduction is the BETTER that gets. It procs when you need it, dodge doesn't always happen when you need it.

I mean... I outline all of this, I'd highly suggest re-reading section 1 and section 6 more in depth.

If you have any more questions after that I welcome you to send me a PM and I can explain it further, especially since I now have experience from both sides of the coin, the healing side and the tanking side.

Kaze has unfortunately missed the point of what I said. I appreciate and understand what you're saying, but I'm not sure you hear me either.

It is not enough to simply say that armor is better than dodge. If you have two pieces of gear, one of which gives a small amount of extra armor, and another which gives a substantial amount of dodge, at some point the item with a large amount of dodge has to better regardless of which stat is supposedly better overall.

Again: It is not enough to simply say that armor is better than dodge. We have to know how much better it is in a quantifiable way so that we can compare the relative value of two items with differing amounts of armor and dodge.

[Tinytankz]

Kaze has unfortunately missed the point of what I said. I appreciate and understand what you're saying, but I'm not sure you hear me either.

It is not enough to simply say that armor is better than dodge. If you have two pieces of gear, one of which gives a small amount of extra armor, and another which gives a substantial amount of dodge, at some point the item with a large amount of dodge has to better regardless of which stat is supposedly better overall.

Again: It is not enough to simply say that armor is better than dodge. We have to know how much better it is in a quantifiable way so that we can compare the relative value of two items with differing amounts of armor and dodge.

In fact Gambling is never better than Armor in Real wow. There has never been nor will be a choice between 50 armor and 150 dodge rating in similar item level tank gear. So the point your making is invalid. In terms of our gearing choices point for point ARMOR IS BETTER THAN DODGE

[Thaurissan]

In fact Gambling is never better than Armor in Real wow. There has never been nor will be a choice between 50 armor and 150 dodge rating in similar item level tank gear. So the point your making is invalid. In terms of our gearing choices point for point ARMOR IS BETTER THAN DODGE

There are not differences that large, but there are differences. Some items have large amount of extra armor. Some have extra dodge rating, and some have none.

The particular case that spawned this discussion for me on another thread is the Glyph of Indomitability vs. the Corpse Tongue Coin. The Glyph provides an extra 1.06% physical damage reduction from armor, the Coin provides an extra 2.26% dodge (on my stat sheet at least, it may vary depending other people's DR situation). That means that on average, over time, the Coin provides over twice as much damage reduction. So how much of a gap does there have to be before a specific amount of dodge is better than a specific amount of armor? That's what we need to know.

[Aggathon]

I honestly do not pay attention to how much avoidance is on gear, I look soley at EHP. The reason is because you cannot stack enough avoidance to push the graph in section 6 into an acceptable RNG (or gambling odds) range for my comfort. I would argue that any amount of EHP is better than anything short of absurd amounts of avoidance in current content. You can't quantify it because it varies depending on your gear and content. I would say that if it is enough dodge to push you into the 75% avoidance + range then it is better, otherwise, stack EHP out da butt.

As time -> infinity odds of dodging enough to prevent a fatal combination -> zero. Period. If you stacked enough avoidance for it to make a noticeable difference in ICC you would sacrifice so much EHP that you'd get 2 shot and the extra avoidance you gained wouldn't matter.

I agree with what they're saying in the other thread, essence of gossimir > CTC. Pretty much ANY trinket with armor or stam on it is better than CTC.

Edit:

Kaze has unfortunately missed the point of what I said. I appreciate and understand what you're saying, but I'm not sure you hear me either.

It is not enough to simply say that armor is better than dodge. If you have two pieces of gear, one of which gives a small amount of extra armor, and another which gives a substantial amount of dodge, at some point the item with a large amount of dodge has to better regardless of which stat is supposedly better overall.

Again: It is not enough to simply say that armor is better than dodge. We have to know how much better it is in a quantifiable way so that we can compare the relative value of two items with differing amounts of armor and dodge.

If you can provide me with alternative theorycrafting to prove me wrong, then by all means I will conceed, but so far you provide NOTHING in terms of solid evidence, only loose (at best) analogies and hypothetical situations.

Prove me wrong and I'll conceed, but here experience and raw math are not in your favor at all.

[Tinytankz]

Interesting point. On average, over time, Those are lots of modifiers there. The problem is your playing a Random Game as i am sure you have been told many times the 1 time the numbers go bad (as in any random environment they can) your looking at a wipe cause you were gambling. Don't think that rolling dice can be relied upon. I would think that at a certain point the levels can play more in your favor but, 2% dodge versus 1% less physical damage is something i cant really look at and go Yeah easy choice. I pick the Certain and guaranteed, Your choice may be different but, It is an argument not worth having. The math Shows Armor Beats Dodge if you find the dodge works in your favor well have fun you lucky monkey.

[Mend]

There are not differences that large, but there are differences. Some items have large amount of extra armor. Some have extra dodge rating, and some have none.

I can't find any situation were you would have the option between a large amount of armor vs. a larger amount of dodge.

These are all the bonus armor pieces http://www.wowhead.com/items=4.4?fil...09;crs=1;crv=0

There are a few situations where you have the choice between larger dodge rating or bonus armor, but the bonus armor is usually much more than what the dodge rating was worth and makes it an unfair comparison.

[Satrina]

Again: It is not enough to simply say that armor is better than dodge. We have to know how much better it is in a quantifiable way so that we can compare the relative value of two items with differing amounts of armor and dodge.

If you can solve the problem of equating the static quantity of EHP with the randomness of avoidance, you'll be a hero to everybody that has ever played an MMO. If not, you're just another in the long line of people who have come along saying there must be SOME way to do it with no actual input on how it can realistically be done.

[Aggathon]

I've bashed my head against a wall for literally days on end. I wasted an entire perfectly good Friday one day working with TGM to come up with some sort of survivability quotient. We couldn't. The best we could come up with is minTTD, minTTDwHeals, and average TTD calculations, which frankly only take into account the tanking damage intake form, only minTTDwHeals and average TTDwHeals matter, but those only come into effect when EHPS < EDTPS, so really you're trying to model the "art" that is healing, and that's damned near impossible to do, might as well play the game and figure it out for yourself, and in my personal experience I almost got benched as a senior officer trying to balance my gear, I went for Raw EHP and became one of the better tanks on the server and at one point my guild was ranked 8th in the US. I'm proof that EHP > RNG.

[Satrina]

Also, because of the way DR is applied to avoidances, you cannot look at a single piece of armour and say "this gives enough extra avoidance to be better than X armour". You have to recalculate your diminished avoidance including the boost from the candidate upgrade piece and use the new avoidance total as your basis to try and do whatever sort of equation with EHP.

Let us know how you make out with the math

[Aggathon]

That's a good point, 2% worth of dodge rating does not = 2% dodge after DRs. There's a table around here somewhere to figure it all out if you use some interpolation.

[Kojiyama]

Well, it is worth noting the the DR of avoidance generally roughly mirrors the increasing returns on the mathematical effectiveness of avoidance--thus canceling out said increasing return properly and yielding a fairly linear increase in survival per rating point.

So, it's a little inaccurate to say that avoidance becomes less valuable due to DR...it becomes roughly as valuable as the previous point due to DR, instead of more valuable as it used to be.

[Thaurissan]

I honestly do not pay attention to how much avoidance is on gear, I look soley at EHP. The reason is because you cannot stack enough avoidance to push the graph in section 6 into an acceptable RNG (or gambling odds) range for my comfort. I would argue that any amount of EHP is better than anything short of absurd amounts of avoidance in current content. You can't quantify it because it varies depending on your gear and content. I would say that if it is enough dodge to push you into the 75% avoidance + range then it is better, otherwise, stack EHP out da butt.

...

If you can provide me with alternative theorycrafting to prove me wrong, then by all means I will conceed, but so far you provide NOTHING in terms of solid evidence, only loose (at best) analogies and hypothetical situations.

Prove me wrong and I'll conceed, but here experience and raw math are not in your favor at all.

But the flip side of what you're saying is that you can't prove your argument either...which means you should not be so determined to promote your argument as irrefutable fact.

With all due respect, since we have conceded that ultimately this can't be modeled and has to be proven out by in-game experience...I have seen the effect you're talking about. And it is a problem that seems to be faced by Warrior tanks to a far greater degree than other classes. They seem to be doing fine and then...splat. They're dead.

Your own experience is echoed by great numbers of Prot Warriors who have posted on various forums about their woes for some time now.

It is entirely possible that with the great number of factors in play, talents like the Paladin's Ardent Defender ameliorate just enough of the worst case scenario that a well-geared Paladin tank rarely dies where a Warrior usually would. Druids, meanwhile, have their huge health pool to rely on, and DKs can self-heal.

It's great that you found a way to make Prot Warrior viable. I'm just not sure it applies to all tanking classes equally.