Elitist Jerks (http://elitistjerks.com/forums.php)
-   Class Mechanics (http://elitistjerks.com/f31/)
-   -   [Tank] Weapon Expertise and Parry Mitigation (http://elitistjerks.com/f31/t18443-tank_weapon_expertise_parry_mitigation/)

 sordee 11/19/07 6:58 PM

[Tank] Weapon Expertise and Parry Mitigation

I had previously posted this on Emmerald's Druid forum, but am looking from some input from a wider audience:
(If this is too narrow a topic please lock)

We all know that Expertise now reduces parries on mobs (and dodges). This is all great and effectively is a "+hit" of a form:

But at the same time with the mob parry mechanic of a faster swing timer (30%?) for the next swing, this is in effect an additional source of Avoidance, correct?

What becomes a good way of valuing expertise in the Mitigation/Avoidance models? Does -1% Parry = +.3% dodge? Or is my math way off?

Quote:
 That is a pretty interesting question. In short, I would actually consider the benefit from reduced parries as "mitigation" instead of "dodge" since it effectively reduces the mob's average swing time. From my perspective mitigation >> avoidance (dodge), so that's a good thing. However, in the modeling world, the two percentages are considered equal. How to quantify it? The tricky thing here is that reduced target parry depends on your attacks. For example, if you never attack, then there are no parries, and the reduction is irrelevant. On the other hand, if you attack quickly (like a rogue, DW warrior, DW shaman, etc.), then there are several opportunities for a parry per swing. How many times do bears attack per second? We have a couple attacks in our tanking cycle. Most are governed by the global cooldown (1.5s GCD) except for maul/white attacks which are on a 2.5s swing timer independent of the GCD. Doing some quick math, the least common multiple of 1.5 and 2.5 is 15. So in 15 seconds we get an integer number of GCD attacks (10) and white attacks (6). That comes to 16/15 attacks per second (1.067). You can safely round that down to 1 attack per second if you consider lag. Now, we need to know how many times the mob attacks so we can calculate the number of opportunities for a parry. I don't have a good reference for this, but I'll use 2.0s as a boss swing timer. I know that's the regular swing timer for Gruul and Prince Malchezar (phase 1, phase 2 is dual wielding 1.5s weapons) because I spent way too much time reading combat logs. This makes the math incredibly easy as we have two of our attacks per one boss attack. In other words, there are two opportunities (on average) for the boss to parry an attack and get the swing modifier. I'm not sure what the parry haste modifier should be. If it's 30%, then I would say that 1% parry reduction = 0.3% * 2 = 0.6% mitigation.
And a few other posters prompted some additional questions:

A) Is the haste modifier 30% as used in these calculations?

B) Are there mobs that don't parry? Also can mobs parry while casting a spell?

C) How much does this vary for a warrior/pally tank using a sub-2.0 speed weapon. (as opposed to the Bear 2.6 speed)

D) Is this all such a small factor, that it isn't worth modeling?

 Xav 11/19/07 7:31 PM

This kind of has been discussed a ton already in the various Warrior/Tanking/Expertise threads on this forum, I'm wondering if you looked through the threads at all before crossposting?

a) I'm not sure what % haste it is after a parry, I thought it was 40, maybe it's 30, maybe it's 50, regardless, it's a lot faster.

b) Bosses can indeed parry while casting a spell, I believe that's the logic behind bosses that turn around for RSTS abilities getting a 'string of parries' from the melee and then owning the MT, or whatever.

c) I don't understand your question. The rate of your attacks directly affects how often you have a chance to get parried. If a warrior is getting off more attacks in a shorter period than a druid, he will have a higher chance of one of them being parried... I don't know what a druid 'tank rotation' is like and how often they have an attack on the boss, so I can't compare it myself.

d) Yes. It's fairly irrelevent, all people should really care about is having only the tank in front of the mob, and everyone else behind. Expertise will reduce the parry->extra attack chances, but there's no real point to look at Expertise as an avoidance stat because of that. Expertise is for threat, above all, and you wear other stuff instead to prevent damage if that's what you're after.

 Aldriana 11/19/07 7:35 PM

The answer is going to depend on a couple factors, including:

1) How often the boss attacks
2) How often the tank attacks, and
3) What the bosses base parry rate is.

I don't really have a good feeling for what all these should be, but making up some numbers, lets say the boss attacks once every 2 seconds, the tank attacks about once per .8 seconds (autoattack with a 1.6 speed weapon and an instant attack every 1.6 seconds as well), and the boss's base parry rate is 10%.

Now, on average, a parry reduces the time between attacks by 24%. Hence, each parry gives an extra .24 of an attack.

If the boss parries 10% of attacks, that means we expect to see a parry every 8 seconds or so. In 8 seconds, the boss attacks 4 times; the parry attacks up this to 4.24. Hence, the 10% parry chance increases the number of incoming attacks by about 6%; Thus, each 1% parry-avoidance should reduce the number of attacks by roughly .6% (making a not-entirely-valid linearity assumption, but as a ballpark figure it should do fine). Thus, the total number of hits against the tank goes down by about 6% as well.

Meanwhile, assuming a tank sitting at, say, 50% avoidance (20% dodge, 20% parry, 10% miss), 1% of direct avoidance will reduce the number of attacks that land by about 2%; hence, 1% expertise grants avoidance comparable to .3% dodge.

Thus, 15.77 expertise rating grants benefit equivalent to roughly 5.7 (18.9*.3) dodge; hence each point of expertise rating is worth about .36 dodge rating.

Note, however, that the parry hasted attacks are probably more dangerous than the average attack, so the real benefit may be larger.

 Meddler 11/19/07 8:16 PM

Quote:
 Originally Posted by Aldriana (Post 554170) Thus, 15.77 expertise rating grants benefit equivalent to roughly 5.7 (18.9*.3) dodge; hence each point of expertise rating is worth about .36 dodge rating. Note, however, that the parry hasted attacks are probably more dangerous than the average attack, so the real benefit may be larger.
Don't forget though that the parry hasted attacks avoided will all only be white attacks. If the boss in question has any special attacks that compose a significant proportion of its damage then dodge will be notably better than the above ratio since it will (usually) give a chance to avoid those whereas expertise won't be helping at all. I realise this is a 'it depends' result but do feel it's important to consider expertise in particular within context.

 Aldriana 11/19/07 9:10 PM

True; there are encounters where special attacks skew this. However, I would argue that such bosses are reasonably rare; for instance, as MT the only encounter in T5+ content that immediately springs to mind in terms of having significant non-autoattack melee damage is Mother Shahraz; there are also a few with some assorted elemental damage thrown in on the top (Kael, Illidari Council, etc.)... but dodge suffers the same deficiencies on those. On the whole, I'd say that the majority of bosses do most if not all of their physical damage via autoattack; hence, with a few notable exceptions, analysis in the style of my previous post should be a reasonably fair estimate.

 Regan_ 11/19/07 10:34 PM

Weapon Expertise, however, mitigates parry burst, and burst kills tanks. If I'm not mistaken, it's still not possible with current itemisation to negate completely a boss parry (I think it's been estimated at 9%?), but is very easy to attain 20-25 expertise with a human/orc tank which negates 5-6% of parry, removing more than 50% of parries a boss will generate during a fight. So at current item price I consider it a very efficient burst mitigation stat.

For the same item budget you can get 1% "antiparry" (which means 11% less parries against a 9% parry chance), you sacrifice only 0,8% dodge. To me, it's a hell of a deal.

 Bullshot 11/19/07 10:49 PM

Quote:
 Originally Posted by sordee (Post 554124) A) Is the haste modifier 30% as used in these calculations?
Quote:
 Originally Posted by Xaviera (Post 554163) a) I'm not sure what % haste it is after a parry, I thought it was 40, maybe it's 30, maybe it's 50, regardless, it's a lot faster.
I was under the impression that a mob attacks instantly after it parries an attack and the 40% haste to the swing timer is only for players who parry an attack. Or am I way off here?

 Aldriana 11/19/07 10:57 PM

My understanding was that bosses use the same mechanics as players for parry haste, namely that it decreases the time to next attack by 40% of weapon speed, to a minimum of 20% of the speed of the weapon; when one actually works out the details of this, one finds that this, on average, increases the speed of the next attack by 24%; hence the number that appears in my calculations above.

 Xerophyte 11/19/07 11:46 PM

Hrm. 24% haste/attack is what you get if you assume a uniform probability distribution over swing time, however this isn't strictly true. With parry potentially hasting through all parts of a mob's swing except the final .2, at any given moment it's more likely that the mob's swing timer is in its final .2 than any other part, meaning that parries are more likely to not do anything than assumed earlier ... or would be, if parries were uniformly distributed in the time frame immediately following another parry, which they're not when you've got a single person in front. 24% is certainly a good approximation though, and I shouldn't be picking nits, especially when I can't be arsed to carry my analysis to a decent enough conclusion..

However, since I'm physiologically incapable of not picking nits I should probably also mention that a mob's typical attack speed is actually 2.4 seconds, thanks to the wonders of Thunderclap. This in turn ups the haste from 10% parry to 7.2% haste and the dodge equivalence to .43 dodge rating per expertise or so. Which is respectable, albeit not exactly mind-blowing.

 Garithras 11/20/07 9:30 AM

I would love to see a good mathematical theorem on this, as I think it's a little bit beyond me. The first part to clarify, however, is getting all the variables as accurate as possible. Most of what we need to know has already been asked: how often bosses parry, how exactly parry works (which leads to how fast a boss will be attacking). The speed a tank attacks would potentially be valuable, but that's adding more complexity than I know how to deal with. The hardest part to quantify is exactly how dangerous a parry is.

At my progression level (first couple bosses in TK & SSC down, about halfway through ZA), still the most likely bosses to instagib a tank are Prince and Gruul, and checking a few random WWS logs I can estimate their parry rate at well over 10% -- my sample size is way too small to be accurate, but Prince was fairly steady at 13%, and Gruul was 16% except for one sample with a mere 7%. Of course since its not my own WWS I don't know what gear they are wearing to throw off results. Mother Shahraz is one I've heard that similarly likes to instagib tanks, and averaging a couple WWS logs of her I can put a rough estimate at 19%. I think we can assume that bosses do not all parry at the same rate (which of course makes calculating expertise value extremely difficult).

According to [Melee Combat] Riddle me this - Parry Mechanics, a parry reduces swing time by 40% unless it would reduce the swing time to less than 20% of your swing speed. It is uncertain as to whether this applies to mobs as well, but for the sake of argument I think we can assume so.

For warriors, the danger inherent with parries beyond the faster attack (or any fast attacks, like Prince's 2 extra hits from flurry) is that it can consume shield block charges fast enough to make you vulnerable to crushes. Avoidance and expertise both help reduce the chances of this occurring -- avoidance by providing decreased chance that the boss will hit the tank consecutively, expertise of course by reducing the chance of the parry happening in the first place.

Here's my probably-incorrect stab at maths. One parry is usually not dangerous. Two in a row or more are where they become difficult for healers to cope with. If we have 50% combined avoidance (defined as that which would completely avoid an incoming hit, miss/parry/dodge) and assume a 15% chance of a boss parrying, which is close to where I am at personally, there is a 7.5% chance (50% * 15%) that a hit will be parried, and the resulting incoming hit will not be avoided. The chance of 2 parries in a row, with neither avoided, is .5625% (7.5% * 7.5%), which puts it into the realm of plausible to not occur once throughout an entire boss fight, albeit perhaps not likely. Calculated with 4 expertise (1% less parry), the chance changes to .49%. Calculated at the maximum expertise available to me (57 rating from [Mallet of the Tides], [Brooch of Deftness] and [Bracers of the Ancient Phalanx], plus the 6 from defiance), a total of 5% less parry, I get .25% chance of 2 consecutive parry + hit.

To get the same results out of avoidance, it requires a fair bit more avoidance than expertise. At 51% avoidance, we get .54% (49% * 15% = 7.35%). To get the same .49% as 1% less parry, it takes an additional ~3% avoidance. To get .25% chance as with 5% less parry, it takes an additional ~16% avoidance.

 Mjollnir 11/20/07 9:40 AM

Quote:
 Originally Posted by sordee (Post 554124) Also can mobs parry while casting a spell?
From the standpoint of mitigation, this doesn't matter. The mob won't counter-attack, as a spell is being cast. Any normal swings must wait until cast complete.
I'd like to claim I've seen parries happen while spells are being cast, but I've no proof to support my claim. I know for a fact dodges do (again no proof), but would assume the mechanics would allow parries too under the circumstances.

 Nite_Moogle 11/20/07 9:48 AM

http://elitistjerks.com/f31/t15257-m...ighlight=parry

The average reduction in weapon speed is 25% on a parry. This isn't the whole story since a parry immediately following a melee swing is a much bigger reduction in weapon speed (up to 40%) and is the usual source of parygibbing, so your reduction of .25% per point is mostly valuable in preventing that from happening moreso than any real tangible amount of mitigation.

 Fellwraith 11/20/07 12:35 PM

All the data I've seen points to a boss parry rate of 14-15%. However, that is affected by encounter mechanics, so you won't see a rate that high in most fights. Trash parry rates are significantly lower (definitely below 8% dodge and 8% parry for level 70-71).

The variation you see in parry rates from one boss to the next is largely from bosses who have a "turn and cast" RSTS effect (bosses like Morogrim or Mother) or have a channeled spell effect (like Rage Winterchill). The RSTS spells are usually instant cast, so they aren't really casting so much as turning to cast. The boss hasn't interrupted his attack timer much (if at all) to use the instant spell, but he is benefitting from parrying a melee dps' attack. The tank cannot be parried at this time, which will make the parry rate look lower than it really is. Bosses won't parry if they're channeling something like death and decay. I've also never seen Kael'thas parry while he's casting a pyroblast or fireball, so it's probably safe to say that anything with a channeling bar and a significant cast time will likely prevent a boss from parrying (fear or breath attacks I'm not as sure about).

As others have already said, expertise is a threat stat, not an avoidance stat. Dodge rating would probably be a more efficient way to avoid damage.

 Garithras 11/20/07 1:09 PM

Quote:
 Originally Posted by Fellwraith (Post 554909) As others have already said, expertise is a threat stat, not an avoidance stat. Dodge rating would probably be a more efficient way to avoid damage.
I won't deny that its most noticeable effect is to increase threat, but if you're going to try and shoot me down please back it up with facts, not hearsay, or at least point to an error in my math or method.

 Napoleond 11/20/07 2:52 PM

Quote:
 Originally Posted by Fellwraith (Post 554909) As others have already said, expertise is a threat stat, not an avoidance stat. Dodge rating would probably be a more efficient way to avoid damage.
It is both, in the fact that:

A. you are getting parried and dodged less, therefore, more hits with abilities = More threat generation.

B. You get parried less, therefore, less thrashes, fast hits by a boss etc. Less hits on you = avoidance.

Its not a huge avoidance stat, as parry rating, dodge rating, and defense will do more to avoid hits, but it does help a tank survive due to unlucky parry strings.

All times are GMT -4. The time now is 5:43 AM.