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 » Tue Sep 08, 2009 7:08 am

Solitatis wrote:Currently i am only worried about three bosses in game: Anub hard, Twins hard, and Algalon. I do not tank adds, have an OT for that :) Hence the question. I guess it is: When is 70% of dodge worth more than pure dodge because of DR? (while still maintaining 1.88 dodge/parry ratio)

The 70% figure was calculated taking diminishing returns into consideration for that gear level.

Using the simulation, it takes 1875ish dodge rating before expertise catches up to dodge. We're not likely to see that gear level this expansion, I think. At least not with a proper 1.88 ratio.

I'll put a disclaimer on this that I'm no longer certain the 70% figure is accurate. I'll go into more detail about that shortly.
"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: 7793
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby Solitatis » Tue Sep 08, 2009 7:25 am

I have also seen two different numbers on "soft cap" 24 from this thread:
http://maintankadin.failsafedesign.com/viewtopic.php?f=6&t=17075

and 26 from yours

Which is correct?
Thought without action is a daydream.
Action without thought is a nightmare.
User avatar
Solitatis
 
Posts: 191
Joined: Tue Apr 15, 2008 6:47 pm

Re: Calculating the avoidance value of expertise

Postby majiben » Tue Sep 08, 2009 7:28 am

26
Amirya wrote:some bizarre lovechild of Hawking, Einstein, and Theck
User avatar
majiben
Moderator
 
Posts: 6999
Joined: Fri Aug 22, 2008 4:37 pm
Location: Retired

Re: Calculating the avoidance value of expertise

Postby theckhd » Tue Sep 08, 2009 7:33 am

There are two issues I'd like some feedback on.
1) I'm not really happy with the Spike DTPS calculations, for a variety of reasons. Cordelia brought up some good points about tank-death scenarios, and I personally am not sure the DTPS*probability product is a very useful metric, in that I'm not sure it does a great job of measuring survivability. I'd like to throw some brainpower at this and see if we can figure out a good metric for tank survivability in spike situations. Something that reflects your chance to prevent death due to a string of melee attacks. Maybe this damage intake / probability product is a decent measure of that, and maybe not, but I think we need to look at the question carefully before we spend any more time refining the numerics.

2) In calculating the weapon speed avoidance for Lythac, I noticed something in the original calculation that may be in error. After calculating parry-haste coefficients and the number of extra attacks the boss gains due to parry-haste, we calculate the amount of expertise it takes to reduce this number by 1 (i.e. to drop from 8 extra attacks to 7). We then convert that to dodge rating and dodge percentage after diminishing returns, and calculate the number of attacks we'd dodge due to the additional amount of dodge rating. We use this to calculate the "attack reduction" effectiveness of expertise rating.

In other words, we're comparing the amount of attacks expertise prevents from occurring (1) to the amount of attacks that extra dodge rating would turn into avoids. This is fine for an "attack reduction" effectiveness, but in reality we would have avoided 60-ish % of the extra attacks that expertise prevented. So I'm not sure whether this metric over-values expertise by a factor of 2 or so, due to that fact.

If we wanted to compare the number of extra attacks that expertise prevents from landing, we would have to multiply the number of attacks prevented by (1-player_avoid), or 0.4ish.

....


....

And now I feel really dumb for typing that, because I noticed that when I calculate the effectiveness %, I compare the equivalent # of attacks dodged to 0.4, not 1, probably for that exact reason. Oops.

Ok, so effectiveness calculation should be correct. Nothing to see here. I'll leave my morning-addled ramblings here just in case someone gets confused by it in the future. Probably me. /sigh
"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: 7793
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby Meloree » Tue Sep 08, 2009 8:27 am

Regarding 1). I don't know that I have the math to keep up with you on this, but lets consider looking at some WotLK bosses where there is (or was during progression) a risk of tank death, and see what parry haste might affect in that scenario.

XT-Hardmode: I'm not sure tank-death was a problem here for anyone, but it's an example of a boss where the tank takes basically straight melee. I believe I got hit for in the neighbourhood of 21k during our hardmode progression on this, on a healthpool of 45k or so. We used one Holy Heal Turret to keep me up, plus whatever splash healing I caught. Tank death was a total nonissue. I don't know if he parry hastes or not, but I didn't have a lot of expertise then. Given that I was solotanking, and therefore taking occassional hits from orbs (8kish nature), I was at risk of being 2shot, but it never seemed to crop up in practice that 2 XT swings and one life spark swing all happenned together so closely no heal could land.

Steelbreaker Last: The primary danger scenario is Fusion Punch -> Melee. I'm not sure if his swing timer is paused or reset during fusion punch cast, and given the ability of all the tank healers to top up the tank fully before fusion punch and time something big to land right after, I'm not sure how relevant the melee swing is. We had tank death problems here, but they were mostly due to one-shots, and/or healer setup, and/or screwing up our cooldown rotation, and/or failure to cleanse the 2nd tank. It's possible if Steelbreaker's swing timer is fully reset that an inopportune parry could give the full 40% haste to his relevant melee swing, and put the healers too far behind post fusion punch, but I'm not sure what the chances are.

Algalon: Our tank death problems were mostly due to healers having to move from cosmic smashes at the same time (bad positioning), even during our pre-nerf attempts. If he has parry-haste enabled, though, this is potentially a fight that makes it scary, he's dual-wielding big hits on a fast timer, with some unavoidable damage to help create spikes.

Mimiron P1: His melee is barely relevant, but during plasma blast I suppose additional damage could cause problems.

Vezax (prenerfs): This is probably the only Ulduar fight where I can see parry haste having massive consequences. The healers are primarily reactive to conserve mana, and he was able to 2-shot a tank at the time. There isn't any damage except melee.

Yogg0: The adds are level 82.

Gormok: Burst scenario is melee/bleed tick/impale/melee. On the other hand, you have 5 or 6 healers who don't have anything else to do in that phase. Still, I assume parry haste could be bad, especially if you're doing one of the fancy 2-tank rotations.

Dreadscale (enraged): Hits for 30-35k, plus burning bile ticks. On the other hand, you have 5 healers with nothing better to do. This was easy to heal through. That said, Melee/Bile/Melee in short order could cause problems. Might be worth simulating.

Anub25: Haven't seen it yet. That said, in normal, the "burst" mechanic on MT is basically "You got frozen", so you're not causing any parry haste there.

So, honestly, the current DPS*probability product doesn't really give a good measure. But I don't know what might be better.

--
Meloree, <Edge>, Garona
Meloree
Maintankadonor
 
Posts: 1420
Joined: Wed Mar 12, 2008 10:15 am

Re: Calculating the avoidance value of expertise

Postby lythac » Tue Sep 08, 2009 8:45 am

Meloree wrote:Steelbreaker Last: The primary danger scenario is Fusion Punch -> Melee.

<Snip>

It's possible if Steelbreaker's swing timer is fully reset that an inopportune parry could give the full 40% haste to his relevant melee swing, and put the healers too far behind post fusion punch, but I'm not sure what the chances are.


Chances would be zero if during the fusion punch cast the tank stops attacking?
Ryshad / Lythac of <Heretic> Nagrand-EU
User avatar
lythac
Moderator
 
Posts: 2675
Joined: Wed Sep 24, 2008 8:10 am

Re: Calculating the avoidance value of expertise

Postby theckhd » Tue Sep 08, 2009 8:49 am

Lythac wrote:Chances would be zero if during the fusion punch cast the tank stops attacking?

Which is actually feasible too, given the aggro lead one should have in this fight.
"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: 7793
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby Meloree » Tue Sep 08, 2009 9:08 am

Lythac wrote:
Meloree wrote:Steelbreaker Last: The primary danger scenario is Fusion Punch -> Melee.

<Snip>

It's possible if Steelbreaker's swing timer is fully reset that an inopportune parry could give the full 40% haste to his relevant melee swing, and put the healers too far behind post fusion punch, but I'm not sure what the chances are.


Chances would be zero if during the fusion punch cast the tank stops attacking?


Okay, cross that one off the list where expertise has the potential to be relevant, then. That said, I don't think that's optimal given how tight the dps check is in progression, given that the tank can easily be doing north of 10k dps with the buff.

--
Meloree, <Edge>, Garona
Meloree
Maintankadonor
 
Posts: 1420
Joined: Wed Mar 12, 2008 10:15 am

Re: Calculating the avoidance value of expertise

Postby lythac » Tue Sep 08, 2009 10:43 am

I was mainly thinking of future use when you know there will be a large attack then a melee after (and threat is not too much of a concern). If you have the chance to reduce parry haste to zero by stoping attacks, then expertise loses its value and you would be better off with avoidance. Sarth 3D maintanking would also come under this category.
Ryshad / Lythac of <Heretic> Nagrand-EU
User avatar
lythac
Moderator
 
Posts: 2675
Joined: Wed Sep 24, 2008 8:10 am

Re: Calculating the avoidance value of expertise

Postby toothdecaykills » Tue Sep 08, 2009 12:10 pm

I'm not exactly on the same level as many of you. I often feel like the child watching all the adults converse at a table I'm still too small to sit at. That's okay, I wait for the distilled left overs to be dumped unceremoniously in my bowl.

Or maybe I'm a dog in this analogy.

Anyway, to me, it seems that anecdotal evidence that nothing bad happened, to reference Meloree, isn't of much use when trying to model the effects of expertise. There's no way of telling, with what little data given, that there existed a parry haste effect that you successully removed or one that happened, could have killed you, but quick healers managed to circumvent. Of course you may ask why we'd care about it if we can't really ever see it working. This isn't really a danger that we can claim we reduced with any sort of certainty. I can't tell you that my expertise, or avoidance, or even my very specific total health, forced a specific event that would have otherwise flattened me. I can however tell when the bad stuff happens, mostly because the results are plain to see. I die. I can then remove the chances of those happening. Those events won't happen much in past content, we've dominated already, but my gear isn't FOR past content. I try to perfect my gear not for the fights I've already won but for those I've yet to win. 

 I guess in that way the absence of evidence is itself evidence. On the hardest encounters that you are the least prepared for, in gear or otherwise, then the risk of parry hasted pwnage is more potent if not directly greater. In this case, it isn't a tank death that simply puts the raid a couple minutes behind farming the night. Instead it's a tank death that directly lowers the ability for your raid to continue learning that encounter. Only in the blisteringly hard progression encounters are considerations like this worth the effort. 
toothdecaykills
 
Posts: 138
Joined: Sat Jul 28, 2007 10:50 pm

Re: Calculating the avoidance value of expertise

Postby Meloree » Tue Sep 08, 2009 1:07 pm

toothdecaykills wrote:Anyway, to me, it seems that anecdotal evidence that nothing bad happened, to reference Meloree, isn't of much use when trying to model the effects of expertise. There's no way of telling, with what little data given, that there existed a parry haste effect that you successully removed or one that happened, could have killed you, but quick healers managed to circumvent. Of course you may ask why we'd care about it if we can't really ever see it working. This isn't really a danger that we can claim we reduced with any sort of certainty. I can't tell you that my expertise, or avoidance, or even my very specific total health, forced a specific event that would have otherwise flattened me. I can however tell when the bad stuff happens, mostly because the results are plain to see. I die. I can then remove the chances of those happening. Those events won't happen much in past content, we've dominated already, but my gear isn't FOR past content. I try to perfect my gear not for the fights I've already won but for those I've yet to win. 


I agree completely, I'm not suggesting that my post be considered as evidence, only the sort of thing to investigate. In other words, those are the types of tank death scenarios we saw in Ulduar and thus far in ToC, it provides a solid basis for beginning to investigate the effects of parry-haste in "real burst" scenarios. Just because I can't recall any parry-hasted burst-gibs during progression doesn't mean they didn't happen, and I apologize if I made it sound like I intended that to be evidence, rather than illustrative.

I don't really have the math to play at this table with regards to modelling this sort of thing, but I think I have the experience to make some contribution regardless.

--
Meloree, <Edge>, Garona
Meloree
Maintankadonor
 
Posts: 1420
Joined: Wed Mar 12, 2008 10:15 am

Re: Calculating the avoidance value of expertise

Postby callimar » Tue Sep 08, 2009 6:00 pm

Interesting stuff... and theck provided mind blowing numbers as usual. However, and I hope its that I just missed it in the swamp of information but, there are two questions that I have. One, how many bosses actually parry haste? If it is a significantly large or small number isn't expertise going to be better or worse based on if the boss will haste? I know right now on some of the 25 HTotC encounters a parry haste is a dead tank. Example on Beasts we use 2 tanks and have tanks getting up to 4 stacks ( just for a split second). Now even backing down to 3 stacks you are looking at 27k ticks aside from boss damage. A parry haste here is a wipe. It was my understanding that because of the relatively small health pools of tanks vs the large incoming damage that most parry hasting had been taken off of bosses. If I am wrong I apologize but I wanted to check on this before I get all excited that expertise is going to provide me avoidance.
callimar
 
Posts: 47
Joined: Thu Jun 14, 2007 10:11 pm

Re: Calculating the avoidance value of expertise

Postby theckhd » Tue Sep 08, 2009 6:13 pm

They turn it off for some bosses, but not all. The majority still have parry-haste enabled. There's a list being compiled over in this thread. Northrend Beasts, for example, all have parry-haste enabled.
"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: 7793
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby callimar » Tue Sep 08, 2009 6:44 pm

Thanks man much appreciated
callimar
 
Posts: 47
Joined: Thu Jun 14, 2007 10:11 pm

Re: Calculating the avoidance value of expertise

Postby Candiru » Wed Sep 09, 2009 1:28 am

http://blue.mmo-champion.com/12/1982032 ... otfix.html

Gormok no longer has parry haste. (hot-fix)
Image
Candiru
 
Posts: 2479
Joined: Mon May 28, 2007 12:21 pm

PreviousNext

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: No registered users and 1 guest


Remove Advertisements

Who is online

In total there is 1 user online :: 0 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: No registered users and 1 guest
?php } else { ?