[Druid]Parries and expertise (or back to the beginning)

bearmount asked:

Here's a question I've been wondering about recently: How adversely does haste affect your survival from additional parry hastes? Basically with all the haste on my new TOC gear I'm wondering if expertise just becomes even more important as a mitigation stat.

One of the very first posts I ever wrote on this blog was about expertise and survival. It's interesting to go back and read that too. The conclusions are sound (stack expertise over avoidance) but it somewhat misses the overall idea, which is whether or not it's actually that important, and whether or not you can circumvent it with simply having more stamina.

Note that it's much less relevant when you can't be crushed. Having two crushing blows in a 40% speed window + a special was much scarier than two normal hits + a special. But that's not what this question is about; this is about how much haste hurts you with parrying.

here's some rough analysis of how haste will affect your survival.
In 240 seconds you have 160 GCDs and 40 cycles of 6 second mangle/x/x/FF. Of those, 3 can be parried. You will also normally have 100 mauls.

So in 240 seconds you have 120 parryable normal attacks (1/4 are faerie fire) and 100 mauls. Let's assume you're soft-capped for expertise, meaning you have 26 expertise. This means that you will have 8.5% parries overall, or 18.7 total parries in the fight.

If you have 1% haste, your attack speed drops from 2.4 to 99% of 2.4, or 2.37. That gives 101 mauls instead of 100, and now you have 221 parryable attacks. You're likely to get 18.78 parries.

How about 10% haste? Your attack speed is 10% faster than normal, or 2.18, and you are expected to have 110 mauls in the same time. Your total parries?

19.5.

So 10% haste means you will, on average, get parried .8 more times than you would have before if you have decent but not stellar expertise in a 4-minute fight.

Can that kill you? Yes, it can. It's not nearly as scary as it was before, but parries can certainly kill you on bosses that parry-haste. Thorim is about the only one I know of that actually has killed anyone recently, but it's theoretically possible.

Should it kill you? Well, even with most hard hitting bosses melee attacks aren't what is going to obliterate you. Thorim is one of the only cases I can think of where it can; most of his damage tends to be scary melee towards the end. Hodir has frozen blows, Vezax isn't really scary because of that at all, Gormok's scary because of impales + melee, Anub is scary because of freezing slash + swarm + melee. Now, if you get unlucky and get two very fast melee hits + a special, you can drop just like that. It's possible.

But I've not seen it happen to me yet.

Why is this the case?

Well, the main reason is the damage profile. Healers have to heal you like they did back in Brutallus days due to how much damage is coming in. They must keep you topped off if at all possible. Because of this, small bursts tend to be wiped out by healing. That's not always the case, but it often is.

Note that before Gormok's parryhaste flag was turned off, he absolutely could parry haste and he absolutely could wipe the floor with someone, especially if they got parried twice in a row. Three quick melee hits within a second is both huge amounts of damage and unexpected for healers to deal with.

Anyway, hope that helps. The long and short of it is that for a pure survival build you should max out expertise if possible and minimize haste if you haven't. If you have, of course, you can safely have as much haste as you want.

[Death Knight] Why dual wielding sucks for tanking

Wait, what? A death knight post? Seriously?

This blog actually started with my wanting to expand arguments I had made several places - most notably at EJ. Seems appropriate that a post I make there spawns my first post on DK tanking here.

Caveat: I don't know anything about DKs, really, other than they're the tank that looks like they belong in a rave when using their abilities. However, one thing I do know a lot about is parry mechanics. Here's the post I made over at EJ:

How we figure out whether a DK is better off using a 2h or 2 1-h comes down to a few things. These would be nice to have, data wise, but don't end up mattering hugely.

What is the typical expertise value of a DK?
What is their typical avoidance %?
What is their typical HP?

They don't matter because either the DK is dealing with content that is trivial (in which case min-maxxing tanking weapons is not important) or the DK is dealing with content that is not (in which case their avoidance, expertise and HP are on the cusp of what is needed to deal with a boss).

After that, we need to take typical tanking weapons into consideration. I would say that a 1.6 speed 1h would be typical (I realize there is a great, slow 1h in naxx, but that is not the standard), and a 3.0 speed 2h weapon would be typical.

The special hit rate between 2h and 1h will remain unmodified. That leaves only white attacks and weapon speed to consider.

With 2 1.6 weapons attacking, you will have 3.75 as many attacks in 3.0 seconds as you would a 2h weapon. This means you have 3.75 times as likely a chance of being parried. Clearly, as you get slower tanking weapons this value goes down. The question is whether the value of using two tanking weapons is worth getting parried almost 4 times as often.

Comparison time. Let's take [Titansteel Destroyer] vs two [Red Sword of Courage].
124 str
105 stam
54 hit
vs
50 str
116 stam
76 defense
52 hit

The gain of 76 defense rating (15.4 defense, or 1.83% total avoidance) is great, but the loss of 74 strength is pretty poor. Still, that's only 18.5 parry rating, or a bit less than .4% parry. So the total avoidance gain is around 1.5%. So to deal with being parried 3.75 times more often in terms of white attacks, you are gaining 1.5% avoidance and about 12 stamina.

To me, this doesn't sound remotely good enough. And note that using the comparison above, it's actually worse; you get parried 4.25 times more often from white attacks.

Finally, note that the overall parries is not 3.75 times as many times due to the special attacks also having parry chances. In a 200 second fight, you would have 133 special attacks (assuming one every gcd, which I admit is fudging things), 67 attacks with a 2h, and 250 attacks with 2 1h. Or 183 more attacks total. That is 1.9 times more than with a 2h alone. So basically you're trading at best 2% avoidance for a doubling of the chance to be parried.

Really, not worth it to me.

Now, what I didn't say at EJ: the effects of a boss parrying you are hard to model, but fairly easy to understand.

When a parry occurs, the target that parried the attack has a chance of having the current swing hasted, as follows:
* If the next attack would normally occur within 20% of your weapon speed after the parry, there is no effect.
* If the next attack would normally occur between 20% and 60% of your weapon speed later, it happens 20% of your weapon speed later instead.
* If the next attack would normally occur more than 60% of your weapon speed later, the time until your next attack is reduced by 40% of your weapon speed.

The overall effect of a parried attack is to speed up a boss's swing timer by roughly 30%. In the worst case it is 40%, but that doesn't happen all the time. ) This means that for using 2 1-handed weapons, you are giving the boss 30% more haste. You have effectively bloodlusted the boss after each parry. 30% more normal swings may not be a lot for someone like Rage Winterchill, but for Patchwerk? And remember, this is not 30% more swings over a nice, long extended time; this is 30% overall more damage that comes in tasty, bursty chunks of death that tends to take your healers off-guard.

Note that I'm not saying doubling the parries makes the boss have bloodlust the entire fight; it depends a lot on how much expertise you have, and I've written about this and parry-gibbing prior to this.

Now, if you're expertise-capped for parry that's something else entirely. Then you'll be trading basically some damage for some avoidance, and that's a great thing to trade. But getting to that 64 expertise value (or 500 expertise rating) is going to be pretty tough, and I bet that you could have just gemmed/enchanted for avoidance and skipped the expertise part anyway.

ETA: one thing I did not factor into consideration is the ability to use two enchants. Having dual goose would help somewhat with the mitigation factor, but even then we're talking about at best 3% total avoidance vs. 30% boss haste. That still doesn't sound like a good deal to me.

[General] The effects of expertise on mitigation

As before, here's the summary: Expertise is the single best stat to stack to reduce the chance of spike damage.

Expertise is understood by most warriors to be their best threat stat. Druids get a lot of good use out of it too. Paladins...well, not as much, threat wise. Since only about 35% of a prot paladin's threat is generated from melee attacks (melee + SoR is basically it) and it doesn't affect taunting, it's not quite as useful as it is to druids or warriors. For druids, all of their attacks are affected by expertise, and expertise is twice as effective as hit rating for threat generation until you reach 24 expertise skill (or 96 expertise rating). It's one of the best if not the best stat to stack for threat.

For Paladins though, it is still pretty useful, and on a per-point basis expertise is almost as good as spelldamage for threat. But that's not the important thing.

For tanks the biggest issue is not how much damage they take per hit; tanking gear tends to normalize to the point where all types of tanks tend to take close to the same damage given the same boss and assuming the same or similar gear levels. What kills a tank is burst damage. In some fights that burst damage can be avoided or otherwise anticipated so that no other damage is taken at the same time; an example would be Magtheridon, where he brings the ceiling down. The raid can stop DPS until after a blast nova, then get him to 30%, cause a lot of raid damage and then resume as normal. In other situations certain bosses will be reduced more by certain classes; Morogrim and Gruul do mostly physical damage, so a druid's high armor comes into play more as a reduction (as an example). Or Prince and a Paladin, where holy shield all but guarantees every attack will be blocked and reduced more. That's one way to reduce spiky damage - make sure that the damage you take is as low as possible.

The other is to reduce the chance of having that spiky damage occur at all. This means we need to understand where spikes come from. One big source is from being parried by a mob. When a parry occurs the mob that parried partially resets their swing timer so that the next attack comes in faster. How much the swing timer is reset or reduced depends on when the parry occurred relative to the swing timer, but in general it's a 40% reduction to the next attack's speed. When this happens, especially in succession, you'll start seeing a lot more spikes of damage - and when this is combined with a boss's special attacks it can quickly lead to too much damage for a healer to overcome. For many fights you can get a parried attack which can lead to two normal melee hits + a special in the space of 2 seconds, and depending on the boss that can be quite a large amount of damage, somewhere on the order of 15-18k. And bosses on average have a 15% parry chance, so this can happen with some frequency. These attacks do enough damage that it's not enough to block them - avoiding them in some way is the only way to reduce that spike. For our purposes, we'll define a spike as being hit by a normal attack, being parried, and then being hit by a normal attack again.

Every 1 point of expertise skill (or 4 points of expertise rating) reduces the chance that you will be parried by .25%. So it takes 4 points of skill or 16 expertise rating to reduce parry chances by 1%. It also takes 19 points of dodge rating to give 1% dodge, which is the best avoidance stat you can stack. So, what are the chances that you will be hit by a normal attack, get parried, and then get hit with another normal attack? The normal attack chance is simply (1-avoidance). The parry chance starts at 15%. Therefore, the chance that you will be parry gibbed is:

(1-avoidance)*(parry chance)*(1-avoidance).

The typical avoidance level for T4-T5 bears and paladins is about 40%. With no expertise at all, you have a 5.4% chance of a parry gib occurring. With each 1% avoidance you reduce that chance by .18%. With each 1% of expertise, however, you reduce that chance by .36%. Thus, expertise is nearly double as valuable as avoidance in stopping a spike from occurring. This pattern holds true for all levels of expertise vs. avoidance:



Admittedly, another great way to deal with spike damage or the chance that you'll take it is to stack health and mitigation so that the spike damage can be more easily dealt with. That is the concept that effective health is built around, and it's always better to maximize your effective health instead of sacrificing it for the purposes of something like threat. That all being said - expertise is the best option for reducing the chances of some types of burst damage killing a tank.