Remove Advertisements

Calculating the avoidance value of expertise

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Re: Calculating the avoidance value of expertise

Postby theckhd » Fri Jan 15, 2010 10:08 am

Awyndel wrote:I don't really want to ruin a whole lot of good theory crafting here, but aren't we misinforming a whole lot of ppl by not mentioning most bosses don't parry haste at all?


"Most bosses" is a bit of an exaggeration. Last I checked, it was only a few isolated hard-hitting slow-swinging bosses. So far on the compiled list, only Gormok, Jaraxxus (still no idea why they chose to do this on him), Ignis, Kologarn, and Patchwerk have it disabled. If anything, "most bosses" have parry-haste enabled, not disabled.

We'll have to do some testing on ICC bosses to see for sure, but it's unlikely they'd turn it off for the newer breed of weaker-hitting faster-swinging bosses. I could maybe see the Gunship captain and Festergut being ones that might have it turned off, since they can both hit pretty hard under the right conditions. Maybe marrowgar too, if his sabre lash is a special ability rather than a regular melee swing.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7762
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby Arees » Fri Jan 15, 2010 10:35 am

theckhd wrote:
Awyndel wrote:I don't really want to ruin a whole lot of good theory crafting here, but aren't we misinforming a whole lot of ppl by not mentioning most bosses don't parry haste at all?


"Most bosses" is a bit of an exaggeration. Last I checked, it was only a few isolated hard-hitting slow-swinging bosses. So far on the compiled list, only Gormok, Jaraxxus (still no idea why they chose to do this on him), Ignis, Kologarn, and Patchwerk have it disabled. If anything, "most bosses" have parry-haste enabled, not disabled.


1) I don't understand why so many people think that bosses don't parry haste anymore. I got into an argument with the warrior tank in my guild about it last week actually. He claimed that they removed parry haste sometime during TBC. The discussion came up because I was talking about how my expertise is starting to go down again and I mentioned this thread and how good expertise is at reducing damage. The argument surprised me because hes a very good tank and reads up on a lot of stuff.

2) I'm pretty sure there was a blue post a little while back saying that Festergut has parry haste disabled.... I'll see if I can find it though.
Image
User avatar
Arees
 
Posts: 546
Joined: Tue Dec 25, 2007 9:50 pm

Re: Calculating the avoidance value of expertise

Postby Hammerjudge » Sat Jan 16, 2010 3:06 am

Thanks Theck, from DK tanks everywhere. Brilliant work.
I also checked it within the limits of my skill, and found no logical problems.

I've posted in the other thread about DW.
Hammerjudge
Maintankadonor
 
Posts: 143
Joined: Mon Oct 01, 2007 8:36 pm

Re: Calculating the avoidance value of expertise

Postby badgermonkey » Sat Jan 16, 2010 8:37 am

theckhd wrote: Maybe marrowgar too, if his sabre lash is a special ability rather than a regular melee swing.



I don't have a link to a log, so this is pure "this is whta I experienced"


But the other day we had 2 tanks tanking Marrowgar, a drood and me. The drood kept randomly dying, and I couldn't be kept up the whole fight alone.

I told the drood to stop autoattacking, since he had like no expertise. To just stand on top of me. The druid stopped taking huge incoming damage.

It's not proof, but it hints at the idea Marrowgar has some Parry-Haste enabled. It was literally changed from druid dropping dead 10 seconds in, to druid survivng the whole fight (he was really undergeared, 32k unbuffed HP in bear form)
Image
badgermonkey
 
Posts: 329
Joined: Tue Apr 28, 2009 6:36 pm

Re: Calculating the avoidance value of expertise

Postby Lieris » Sat Jan 16, 2010 5:11 pm

Is 10exp/15stam a good alternative to 10agi/15stam (assuming beyond the soft cap) as our red gem? I'm leaning towards putting the exp hybrid gem into the badge belt tomorrow based on what I've read in the OP.
Lieris
Maintankadonor
 
Posts: 2181
Joined: Mon Jun 18, 2007 7:49 am

Re: Calculating the avoidance value of expertise

Postby theckhd » Sun Jan 17, 2010 4:02 pm

Lieris wrote:Is 10exp/15stam a good alternative to 10agi/15stam (assuming beyond the soft cap) as our red gem? I'm leaning towards putting the exp hybrid gem into the badge belt tomorrow based on what I've read in the OP.

I think I'd still go with the agility/stam gem, personally. It's slightly less threat, but gives some EH via armor.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7762
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby Thels » Mon Jan 18, 2010 7:23 am

badgermonkey wrote:I don't have a link to a log, so this is pure "this is whta I experienced"

But the other day we had 2 tanks tanking Marrowgar, a drood and me. The drood kept randomly dying, and I couldn't be kept up the whole fight alone.

I told the drood to stop autoattacking, since he had like no expertise. To just stand on top of me. The druid stopped taking huge incoming damage.

It's not proof, but it hints at the idea Marrowgar has some Parry-Haste enabled. It was literally changed from druid dropping dead 10 seconds in, to druid survivng the whole fight (he was really undergeared, 32k unbuffed HP in bear form)


Either you or me is thinking wrong here somewhere.

I'm assuming that, since the druid kept dying all the time, you were maintanking, and the druid was only there to soak the saber lashes. (If the druid was actually MTing before the try where he stopped attacking, consider this a non-issue.)

Since the Druid was OTing, all the damage he should be getting were the Saber Lashes (and whirlwind/frostfire where applicable). Even if his attacks were parried a lot, it would be the MT, not himself, who would be eating the extra hits, and therefor not causing him to die more often (unless this substantially increased the damage the MT took, thereby drawing the healers' attention away from the druid).
ImageImage
User avatar
Thels
 
Posts: 1264
Joined: Tue May 15, 2007 7:30 am
Location: The Netherlands

Re: Calculating the avoidance value of expertise

Postby badgermonkey » Mon Jan 18, 2010 7:27 pm

Thels wrote:
Either you or me is thinking wrong here somewhere.

I'm assuming that, since the druid kept dying all the time, you were maintanking, and the druid was only there to soak the saber lashes. (If the druid was actually MTing before the try where he stopped attacking, consider this a non-issue.)

Since the Druid was OTing, all the damage he should be getting were the Saber Lashes (and whirlwind/frostfire where applicable). Even if his attacks were parried a lot, it would be the MT, not himself, who would be eating the extra hits, and therefor not causing him to die more often (unless this substantially increased the damage the MT took, thereby drawing the healers' attention away from the druid).


It could be me who's wrong then. I was under the impression the Saberlash coincided with his regular melee swing. If so then I think my idea might still stand. If not then the druid was dying because his gear or healers suck, and nothing to do with parry-hasting, and we got lucky when he stopped DPSing :P
Image
badgermonkey
 
Posts: 329
Joined: Tue Apr 28, 2009 6:36 pm

Re: Calculating the avoidance value of expertise

Postby Cliffton » Thu Jan 21, 2010 12:55 pm

I approached the question of "how important is expertise?" from a different angle and I think the community in this thread would be interested in my answer.

theckhd wrote:The whole point of the calculation was to show that by trading X dodge rating for X expertise rating, you retain ~80% of the incoming damage reduction and gain more effective reduction of spike damage intake (generally considered more dangerous).


I looked more at the spike damage intake by taking a typical boss swing speed and plotted a moving average for a couple of different window sizes.

Cliffton (past self) wrote:I tend to get parried 1-7 times while tanking a typical boss fight. These bosses have about a 14% chance to parry, and I have about 11% of that reduced through expertise. The charts below postulate what the incoming damage on a tank from boss autoattacks looks like during these parries.

The boss is assumed to autoattack every 2 seconds for 20k damage (typical).

I considered two scenarios. In the first scenario (left side of plot around 10 seconds), I considered what the boss's dps output looks like if a single attack were parried. I generated a few data series containing the moving average of incoming dps for a variety of moving average window sizes. This is the typical damage spike that would be seen when a tank's attack is parried. In the second scenario (right side of plot around 25 seconds), I considered a bad case such that a single boss swing is parry hasted twice.

In a typical fight, if I'm parried 7 times, that means that there are 7 occassions during the fight in which I give the boss an opportunity to cause the first scenario damage spike to occur. Notice that when the boss does not parry, the incoming damage goes up and down predictably and periodically. However, when a parry occurs, the "valleys" disappear and this sets the tank up for other failure scenarios.
pgib0.jpg
"Parry Gib" Damage Spike
pgib0.jpg (101.57 KiB) Viewed 1527 times


I worked pretty much independently from this community and second the conclusion that Expertise is more valuable than it appears at first glance as a tanking stat. My philosophy is that tanks typically die when a bunch of bad things happen at once to the tank. Reducing the frequency with which bad things can happen to the tank is important to failsafe design. Parry haste damage spike is one bad thing that, while on its own probably won't kill a tank, cumulatively with all the other bad things that could happen at the same time, could be a very significant contributing factor.

Taking the broad statement quoted above that expertise is about 80% of the incoming damage reduction that parry/dodge would be, in conjunction with parry damage spikes contributing to horrible messy deaths, leads me to speculate that expertise as a parry-hastable raid boss tank, until hard cap, is more important than dodge or parry.

In other words, I wouldn't set expertise rating equal to 0.8 value of dodge and parry rating in my rawr config. I would set it above parry and dodge - somewhere between 3rd and 5th most important tanking stat behind Stam and Armor.
Cliffton
 
Posts: 17
Joined: Thu Jan 21, 2010 12:19 pm

Re: Calculating the avoidance value of expertise

Postby Fridmarr » Thu Jan 21, 2010 1:31 pm

But both dodge and parry also reduce burst frequency (consecutive hits), and do so quite dramatically with the more you have. At 50% avoidance the odds of a 3 hit chain landing are only 12.5%, and it would cut the average number of parry hasted attacks that land by half on average as well.

EDIT:
Also, don't forget that quite a few bosses, especially the ones that can hit really hard, have the parry haste flag turned off.
Fridmarr
Global Mod
 
Posts: 9666
Joined: Sun Apr 08, 2007 1:03 am

Re: Calculating the avoidance value of expertise

Postby Roots » Thu Jan 21, 2010 1:59 pm

Fridmarr wrote:But both dodge and parry also reduce burst frequency (consecutive hits), and do so quite dramatically with the more you have. At 50% avoidance the odds of a 3 hit chain landing are only 12.5%, and it would cut the average number of parry hasted attacks that land by half on average as well.

EDIT:
Also, don't forget that quite a few bosses, especially the ones that can hit really hard, have the parry haste flag turned off.


Consider the case of specials though eg. where 2 attacks + 1 unavoidable attack become the dangerous situation, and dodge's value is reduced in relation.
Image
User avatar
Roots
 
Posts: 320
Joined: Tue Aug 11, 2009 1:59 pm

Re: Calculating the avoidance value of expertise

Postby theckhd » Thu Jan 21, 2010 2:30 pm

Cliffton wrote:I approached the question of "how important is expertise?" from a different angle and I think the community in this thread would be interested in my answer.

I looked more at the spike damage intake by taking a typical boss swing speed and plotted a moving average for a couple of different window sizes.


I made a very similar calculation (in concept) of spike damage production earlier in the thread. It basically compared the incoming DTPS in different scenarios and used the probability of such a scenario to calculate a "spike DTPS-probability product" of sorts.

We decided later on that while the intuition is correct, we didn't feel confident that the metric had any merit. The discussion of the validity of the metric starts around here.

So your demonstration and my calculation agree qualitatively. But it's hard to put an actual number on it without a true quantitative metric. Is it twice as good? Three times? One and a half? What do you use as a basis for the value?

Coming up with a meaningful metric is a discussion I'd like to resume at some point, if we can come up with some good ideas.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7762
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby Cliffton » Thu Jan 21, 2010 7:50 pm

Fridmarr wrote:But both dodge and parry also reduce burst frequency (consecutive hits), and do so quite dramatically with the more you have. At 50% avoidance the odds of a 3 hit chain landing are only 12.5%, and it would cut the average number of parry hasted attacks that land by half on average as well.
The advantage that an expertise capped tank has is that they would see a more forgiving worst-case damage spike. My analysis is focused on looking at the worst-case damage spikes that may occur, i.e. "If parried, then the tank may potentially be punished by this much additional burst damage taken." During expected scenarios, such as strings of multiple hits in a row, the tank's raid should have no problem keeping the tank alive. It is the unhappy scenarios in which the additional burst damage potential is realized that the raid may struggle to keep the tank alive through.

Depending on class, talents, glyphs, etc., it is typically a tradeoff of about 5 to 10% parry/dodge to reach expertise cap.

Theckhd wrote:Coming up with a meaningful metric is a discussion I'd like to resume at some point, if we can come up with some good ideas.

I'll kick this briefly and see if it snowballs. I would begin defining a quantitative metric by identifying the terms that compose such a metric. Tank's expertise benefits a raid in the following ways, not necessarily in this order:
1. Reduction in peak worst-case DTPS on the tank from parry-haste
2. Reduction in total damage taken by the tank from loss of parry-haste
3. Steadier threat generation at the first few seconds of a fight, thus allowing dps to reliably push out max dps earlier, thus making the encounter shorter, thus rolling the dice fewer times to see if the tank dies.
4. Greater dps/tps by the tank, thus marginally shorter fight.

I would guess that 1 and 2 are the most important benefits that a tank would see from expertise by an order of magnitude, but terms 3 and 4 should be demonstrated negligible before being neglected outright if the goal is to construct a meaningful, precise metric that captures the value of tank expertise to a raid.
Cliffton
 
Posts: 17
Joined: Thu Jan 21, 2010 12:19 pm

Re: Calculating the avoidance value of expertise

Postby Fridmarr » Thu Jan 21, 2010 8:13 pm

Cliffton wrote:
Fridmarr wrote:But both dodge and parry also reduce burst frequency (consecutive hits), and do so quite dramatically with the more you have. At 50% avoidance the odds of a 3 hit chain landing are only 12.5%, and it would cut the average number of parry hasted attacks that land by half on average as well.
The advantage that an expertise capped tank has is that they would see a more forgiving worst-case damage spike. My analysis is focused on looking at the worst-case damage spikes that may occur, i.e. "If parried, then the tank may potentially be punished by this much additional burst damage taken." During expected scenarios, such as strings of multiple hits in a row, the tank's raid should have no problem keeping the tank alive. It is the unhappy scenarios in which the additional burst damage potential is realized that the raid may struggle to keep the tank alive through.
With the removal of crushing blows, burst is often caused by consecutive hits. Unhappy time also includes healers unable to heal for one reason or another, which can easily span multiple hits. Yes, the worst case spike would be less, but I don't find that to be enough of a reason to increase its value by 25%, especially when the ability of avoidance to reduce burst isn't being considered at all. You could possibly make that argument for the final points to cap you, but until then the worst case is the same.
Fridmarr
Global Mod
 
Posts: 9666
Joined: Sun Apr 08, 2007 1:03 am

Re: Calculating the avoidance value of expertise

Postby Cliffton » Thu Jan 21, 2010 8:51 pm

The probability of 3 consecutive hits does not change wildly between the tank who chooses expertise from the tank who chooses dodge. If a player had to sack 10% to cap expertise, the chances change from 12.5% for the tank with 50% avoidance to 21.6% for the tank with 40% avoidance - for 3 hits. The probability of the event of 3 consecutive hits occuring becomes pretty high over the duration of typical encounters, high enough that it occurs multiple times.

Lemma: 3 consecutive hits is likely to happen often during a typical encounter, with or without expertise leeching away parry & dodge rating.

The raid healers must be prepared to deal with this on a regular basis. It's not a shock to them if the tank gets hit a few times in a row. It's business as normal for them regardless of the tank's gear strat.

I think tanks should make it one of their goals to bore healers to death with business as normal healing requirements.
Cliffton
 
Posts: 17
Joined: Thu Jan 21, 2010 12:19 pm

PreviousNext

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: Google [Bot] and 1 guest


Remove Advertisements

Who is online

In total there are 2 users online :: 1 registered, 0 hidden and 1 guest (based on users active over the past 5 minutes)
Most users ever online was 380 on Tue Oct 14, 2008 6:28 pm

Users browsing this forum: Google [Bot] and 1 guest
?php } else { ?