Remove Advertisements

Calculating the avoidance value of expertise

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Re: Calculating the avoidance value of expertise

Postby cordelia » Fri Aug 28, 2009 3:38 pm

Expertise as avoidance is a fascinating concept inasmuch as its intrinsic mechanic is entirely different than the other traditional avoidance stats.

Tanking theorycraft tends to refer to two main benefits that avoidance provides:

(1) Decreasing average damage intake over the course of a fight, thereby relieving the load on healers.

(2) Decreasing the probability of a string of hit-heavy attacks that will insta-gib the tank.

In evaluating expertise's value towards benefit (1), your calculations are perfect. The only statistic that matters is the reduced average number of hits taken by the tank over the long run. The tank takes less damage over the course of the fight and thereby needs to receive fewer heals, saving mana. However, the value of (1) is oft-debated among tanks and healers. On damage heavy fights, some argue that (1) merely increases the percentage of over-healing and saves no mana. Expertise's equivalence to traditional avoidance with regards to this analysis, however, is correct.

In evaluating expertise's value towards benefit (2), things get tricky. I don't have a strong intuition on whether expertise or traditional avoidance is better at (2). Expertise obviously decreases the probability of a parry-gib. However, expertise does nothing to reduce the probability that the tank will be hit by 4 straight melee hits, parry-hasted or not.

Ulduar tanks now stack enough EH to take 2 consecutive melee hits. Assuming the tank is consistently topped off at the start of our string of hits, we must address the probability of X melee hits of size, m, over a span, T, of time. A tank will die when X*m > HP + HPS*T, where HP is the HP of the tank, and HPS is the incoming Healing per Second. Traditional avoidance decreases the expected value of X, decreasing the probability of the first term being high, whereas expertise increases the expected value of T, increasing the value of the second term.

I've long held that the optimum survivability rating for tanks to be looking at is neither EH nor avoidance, but rather a rating derived from a probability function based on expected number of hits occurring over a set amount of time with two types of incoming damage (melee/spike and steady) and a function for incoming heals over time.

TLDR: Expertise as avoidance is a fascinating concept and I don't have a strong enough grasp yet on how effective it is vs. how effective traditional avoidance is at increasing survivability during spike damage.

PS> FWIW, I did napkin math regarding this when expertise first came out in TBC, and agreed that it was significant. It's much greater now, however with double-dib SoV, JoV, and HotR.

PPS> Did you include the extra parryable SoV application from HotR in your calculations? shouldn't be a big boost, but it would scoot expertise's value up a couple percentage points, I think.
User avatar
cordelia
 
Posts: 630
Joined: Tue Oct 09, 2007 3:33 am

Re: Calculating the avoidance value of expertise

Postby Doxa » Fri Aug 28, 2009 4:46 pm

cordelia wrote:It's much greater now, however with double-dib SoV, JoV, and HotR.


Double-dib?

Strange. I'm assuming you're not from 'round here. We say "double-dip".
Tanking Trash on Doxa with Consecration(Rank3) Since 2007, on Doxah with Thunderclap since 2009 and on Doxeh with Death and Decay since 2010

Meanwhile, pulling DPS from fire with life-grip on Enlight and throwing lightning on Pneuma.
Doxa
Maintankadonor
 
Posts: 483
Joined: Fri Jun 13, 2008 9:03 am

Re: Calculating the avoidance value of expertise

Postby theckhd » Fri Aug 28, 2009 5:56 pm

cordelia wrote:In evaluating expertise's value towards benefit (2), things get tricky. I don't have a strong intuition on whether expertise or traditional avoidance is better at (2). Expertise obviously decreases the probability of a parry-gib. However, expertise does nothing to reduce the probability that the tank will be hit by 4 straight melee hits, parry-hasted or not.

Ulduar tanks now stack enough EH to take 2 consecutive melee hits. Assuming the tank is consistently topped off at the start of our string of hits, we must address the probability of X melee hits of size, m, over a span, T, of time. A tank will die when X*m > HP + HPS*T, where HP is the HP of the tank, and HPS is the incoming Healing per Second. Traditional avoidance decreases the expected value of X, decreasing the probability of the first term being high, whereas expertise increases the expected value of T, increasing the value of the second term.

I haven't run the math on this, and I agree it's tricky. I think your approach is correct though, we'd want to know the probability of X melee hits over time T. Traditional avoidance does this by reducing X (because we dodge some of them), in the fashion X=X0*(1-avoidance), where X0 is the base number of avoidable attacks you'd take in time T. You could add a term Y for attacks that are unavoidable, or subject to only certain types of avoidance, but let's not add that complication now.

However, you could also treat expertise as reducing X instead of reducing T. Keep T constant, but modify the expression for X to have a parry-haste term in there to represent those extra attacks due to parries: X=X0*PH*(1-avoid). So expertise reduces X by reducing PH (to a minimum of 1). This should be entirely equivalent to treating X ans constant and letting expertise affect T (which makes sense - you could get there with a simple rearrangement of your inequality and a simple change of variables).

Either way, you're right that it won't reduce the chance of 4 consecutive un-parry-hasted melee hits, but it does reduce the chance of that 4-string happening even faster than T, so it has to come into the probability function somewhere.

cordelia wrote:PPS> Did you include the extra parryable SoV application from HotR in your calculations? shouldn't be a big boost, but it would scoot expertise's value up a couple percentage points, I think.

Yup, it's in the formula for player_attack_speed, in the form of a (1-avoid)/6 term. The HV applications from auto-attacks are there as well (similar term divided by player_swing_speed rather than 6).
"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: 7717
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Calculating the avoidance value of expertise

Postby Robbert » Fri Aug 28, 2009 7:19 pm

cordelia wrote:In evaluating expertise's value towards benefit (2), things get tricky. I don't have a strong intuition on whether expertise or traditional avoidance is better at (2). Expertise obviously decreases the probability of a parry-gib. However, expertise does nothing to reduce the probability that the tank will be hit by 4 straight melee hits, parry-hasted or not.


The way I would look at this point is very much in line with the old EH vs avoidance discussions going back in BC on these forums.

One of the key concepts of the Avoidance approach was it resulted in a reduction in the probability that a worst case scenario (to use the current example) of 4 consecutive melee hits occurred for any given 4 consecutive melee attacks. As avoidance goes up the average incoming damage goes down.

One of the key concepts of the EH approach was being able to survive the worst case scenario under the assumption that it would happen and had to be a survivable scenario. You could not get enough avoidance to guarantee you'd never take 4 straight melee hits, only reduce the probability that it will occur on any string of 4 melee attacks.

When looking at expertise, it occupies a very interesting middle ground. As expertise goes up the probability that any given melee attack coming in will be melee hasted due to the tank being parried, and at this point it's basically the same as the avoidance approach where you're not eliminating the chance of a parry gib, only reducing the probability. Once you reach the parry cap on expertise, however, it moves to basically the same concept as the EH approach where you're mitigating the worst case scenario into a survivable scenario by completely eliminating the chance of a parry gib due to the tank's attacks.

To specifically address your point, you're correct in that expertise does absolutely nothing to reduce the probability of those 4 melee hits landing. In that same case expertise does reduce, or when parry capped completely eliminate, the possibility of one or more of those attacks being parry hasted. In that respect, expertise does serve to reduce or eliminate the worst case scenario. Personally, I'll gladly lose out on a bit of 'real' avoidance in order to reduce or eliminate that worst case scenario from occurring, even if that means not quite as worst case scenario is likely to occur more often over time.
Image
Robbert
 
Posts: 267
Joined: Tue Oct 02, 2007 2:34 pm

Re: Calculating the avoidance value of expertise

Postby Rhiannon » Fri Aug 28, 2009 7:35 pm

theckhd wrote:Also, having just a beacon is dangerous because if nobody takes enough damage to warrant a Holy Light, and he switches to FoL, your incoming HPS might drop enough to kill you. But that's a different point entirely.


With the change to beacon mechanics (beacon transferring overheals now) last patch that'd be a pretty silly thing for him to do if he knows that your surival depends on his beacon heals though!
Rhiannon
 
Posts: 1061
Joined: Mon Mar 03, 2008 8:17 am

Re: Calculating the avoidance value of expertise

Postby Awyndel » Fri Aug 28, 2009 7:59 pm

Ok really sorry I didnt read ( or understand ) all of this, so this might be really stupid questions. Also having a bit trouble with how I should put this.

Theck did you take into account that the spike damage expertise prevents is really only a small percentage of the overall spike damage we get? Did you calculate it's relative use compared to a situation with less expertise or it's real overall use?

Did you also take into account that even if a parry hasted attack is triggered it can still be avoided with normal avoidance when comparing values of avoidance vs expertise.

I heard parry haste was a thing of the past. Wich bosses actually do or don't have this.

Sorry I am really tired now I think I'm gonna sleep and look at this in the morning see if it makes sence.
User avatar
Awyndel
 
Posts: 672
Joined: Sat Feb 14, 2009 8:49 am
Location: The Netherlands

Re: Calculating the avoidance value of expertise

Postby cordelia » Sat Aug 29, 2009 12:17 am

theckhd wrote:I haven't run the math on this, and I agree it's tricky. I think your approach is correct though, we'd want to know the probability of X melee hits over time T. Traditional avoidance does this by reducing X (because we dodge some of them), in the fashion X=X0*(1-avoidance), where X0 is the base number of avoidable attacks you'd take in time T. You could add a term Y for attacks that are unavoidable, or subject to only certain types of avoidance, but let's not add that complication now.

This, btw, has been one of my long-term projects that I started, oh, 2 years ago in Matlab, but never finished. I wanted to write a sim to figure out the EH/avoidance sweet spot, given the size, frequency of boss hits, hps, and spell damage.

However, you could also treat expertise as reducing X instead of reducing T. Keep T constant, but modify the expression for X to have a parry-haste term in there to represent those extra attacks due to parries: X=X0*PH*(1-avoid). So expertise reduces X by reducing PH (to a minimum of 1). This should be entirely equivalent to treating X ans constant and letting expertise affect T (which makes sense - you could get there with a simple rearrangement of your inequality and a simple change of variables).


Very clever. When talking about expected values I entirely agree. I'm not 100% comfortable with the change once we start talking about survivability, though. Hits are binomial, so they either hit or they don't, and insofar as survivability, that makes a difference. On the other hand, we're looking at overall probabilities, so perhaps using the expected value of X from parry haste wouldn't skew the final numbers. Dunno. Something to ponder.

Robbert wrote:The way I would look at this point is very much in line with the old EH vs avoidance discussions going back in BC on these forums.


The issue is, IMO, that the old EH vs. avoidance debate is a false dichotomy. Worst Case Scenario can be measured and expressed in terms of a probability. Your job as a tank is to minimize the probability of death. This can be calculated given the terms of boss melee size, boss swing speed, other incoming damage, armor, health, avoidance, block and block rating, incoming heals and healer reaction to spike damage. The fundamental tenet of EH was to gear for the worst case scenario to ensure survivability. What I'm saying is that optimum survivability comes from finding that sweet spot between EH and avoidance that minimizes the probability of tank death occurring. The worst case scenario isn't X hits in a row. It is X hits of size m occurring over a time t with ongoing damage d and incoming heals H resulting in tank death. I'm not talking about minimizing the chance of 3 hits in a row(traditional avoidance argument), I'm talking about minimizing the probability of all the different combinations of hits that could occur that could kill the tank over the course of the whole fight.

Expertise is similar. It's not one or the other, it's some middle ground who's eventual shape can only be determined by studying the probability functions involved with survivability. You have a gut feeling that eliminating that chance for parry-hasted gibbing is better than eliminating the chance for consecutive blows. Honestly, I think you may be right. But it's a complicated function, and the ratio of which one is better, and by how much isn't known yet.
User avatar
cordelia
 
Posts: 630
Joined: Tue Oct 09, 2007 3:33 am

Re: Calculating the avoidance value of expertise

Postby Awyndel » Sat Aug 29, 2009 6:18 am

Awyndel wrote:Ok really sorry I didnt read ( or understand ) all of this, so this might be really stupid questions. Also having a bit trouble with how I should put this.

Theck did you take into account that the spike damage expertise prevents is really only a small percentage of the overall spike damage we get? Did you calculate it's relative use compared to a situation with less expertise or it's real overall use?

Did you also take into account that even if a parry hasted attack is triggered it can still be avoided with normal avoidance when comparing values of avoidance vs expertise.

I heard parry haste was a thing of the past. Wich bosses actually do or don't have this.

Sorry I am really tired now I think I'm gonna sleep and look at this in the morning see if it makes sence.


Slept on it, but it still makes sence to me. Let's say we're talking about preventing 2 hits in a row here, since that is what a parry haste will mainly cause. Now let's say you have 62% avoidance raid buffed, not at all a weird thought in current gear with libram proc up. Now excuse my newby math but here goes. Let's say 0.38x0.38 so about 14.4% of the incoming attacks will be followed by a second one. Now you said in a 5 min fight we get 4.2 parry hasted attacks ( not sure if that's still up to date for your current code though ), so with 2.0 swing timer that's what, 2.8%. So you can see here wich source of 2 hits in a row is bigger. Now if we start talking about 3 or 4 hits in a row, stuff that kills us, avoidance might get further ahead, although I am not sure coz a parry hasted attack can also be part of longer spikes as well.

Secondly let's not forget that in this scenario 62% of the parry hasted attacks are being avoided by standard avoidance. So if you'd take away 62% of those 2.8% your left with a lot less. I know it doesn't work like that when trying to see what expertise prevents but it is something important to think of when making comparisons between the value of the 2.

I am also very interested in how effective using a slow weapon would be for this. Even a dps weapon at that. Let's say a 2.6 speed weapon. Wouldn't that take at least 30% of the parries away in a fight? Great for threat as well. Although 0.3 of 2.8%x0.32 is maybe just about the same as the avoidance on a tank weapon provides, not sure.

Then there is the rumour that blizz is making it a trend for hard hitters to not parry haste. Not sure how many of those are confirmed though.

I will honesly admit I can't copy the math, but my feeling tells me dodge/parry will always rule over expertise by a long shot. Dodge/parry as 1 and expertise as 0.7 would give an expertise itemized item a slight benefit over a def/dodge/parry item, but I don't believe the numbers are like that when it comes to preventing spikes.

cordelia wrote:The issue is, IMO, that the old EH vs. avoidance debate is a false dichotomy. Worst Case Scenario can be measured and expressed in terms of a probability. Your job as a tank is to minimize the probability of death. This can be calculated given the terms of boss melee size, boss swing speed, other incoming damage, armor, health, avoidance, block and block rating, incoming heals and healer reaction to spike damage. The fundamental tenet of EH was to gear for the worst case scenario to ensure survivability. What I'm saying is that optimum survivability comes from finding that sweet spot between EH and avoidance that minimizes the probability of tank death occurring. The worst case scenario isn't X hits in a row. It is X hits of size m occurring over a time t with ongoing damage d and incoming heals H resulting in tank death. I'm not talking about minimizing the chance of 3 hits in a row(traditional avoidance argument), I'm talking about minimizing the probability of all the different combinations of hits that could occur that could kill the tank over the course of the whole fight.


Absolutely agreed. I have been looking forward for such a simulation for a long time, sometimes bugged theck about it. Unfortunately I dont have the time or skills to do one. Do keep in mind this will all be sensitive to how healers perform. But one could take that into account by simply not taking constant healing for granted in the simulation. We do have to put this in perspective of current items though, as always. And that gives of a very good hint of what to gear for for hard hitters. And I have good feeling that the odds wont be bad on soft hitters then either.
Last edited by Awyndel on Sat Aug 29, 2009 8:25 am, edited 2 times in total.
User avatar
Awyndel
 
Posts: 672
Joined: Sat Feb 14, 2009 8:49 am
Location: The Netherlands

Re: Calculating the avoidance value of expertise

Postby Meloree » Sat Aug 29, 2009 7:17 am

Awyndel wrote:Do keep in mind this will all be sensitive to how healers perform. But one could take that into account by simply not taking constant healing for granted in the simulation.


I would argue that any simulation involving constant healing is inherently flawed. If we look at XT hardmode, for example, he's a pretty light hitter by hardmode standards, hits for in the ballpark of 20k, roughly every 2 seconds. My assigned healing is a beacon. Said paladin knows he's primarily responsible for me, he pays attention. Any druids in the raid also tend to maintain a rejuv, just in case the holy paladin has to move from bomb, so they can swiftmend if they see me drop below 20k. Any shamans in the raid typically have earthshield on me, for lack of a better place to put it. And that's the absolute simplest fight to simulate, the boss doesn't do anything else to the tank that would cause the healers to react. Compare to Algalon with a 1s swing timer, dual wielding for 20kmh, 10k oh, plus 15k from black hole explosions every few seconds, phase punches, quantum strikes, and potential cosmic smashes. We use 2 holy paladins, and a disc priest that raid heals but puts penance into the tanks on cooldown. The only real "oh crap" scenario with that setup is both holy paladins having to move from cosmic smash at the same time.

A group of healers in a raid has nearly unlimited tools to save a tank if they know a spike is coming, or if the spike can be drawn out long enough to enable them to react. And any simulation that doesn't include reaction from a group of healers would generate a flawed analysis, in my opinion. Constant Stream of Healing, though, toss it out the window before you start, it doesn't match reality.

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

Re: Calculating the avoidance value of expertise

Postby cordelia » Sat Aug 29, 2009 11:12 am

Meloree wrote:I would argue that any simulation involving constant healing is inherently flawed.


No one's arguing the other case with you. Incoming healing should be modeled as a function of time. HPS(t), where t is the time elapsed from the start of spike damage. HPS(t) is obviously an increasing function, but will likely cap out at the max throughput of the MT healers. So HPS(0) is the sustained "maintenance" HPS of MT healing, and HPS(6) would be max throughput.

Meloree wrote:A group of healers in a raid has nearly unlimited tools to save a tank if they know a spike is coming, or if the spike can be drawn out long enough to enable them to react.


I'm unconvinced of this. 25k melee hits landing every 2 seconds requires close to 13k HPS to heal through. Add in some dots or AoE damage, and you probably need around 16-17k HPS. I've seen lots of tank deaths because of sustained 4-6 seconds of spike damage. The initial spike was healed up, but the prolonged incoming damage out-damaged the increased incoming heals. Remember that when you assign 2 healers to MT, and the rest to the raid, when raid healing is necessary, you can only rely on the emergency buttons and incoming heals of the two MT healers. Furthermore, sometimes the oh-shit healing cd's are down, NS, penance, holy shock. That slows down max incoming HPS quite a lot.
User avatar
cordelia
 
Posts: 630
Joined: Tue Oct 09, 2007 3:33 am

Re: Calculating the avoidance value of expertise

Postby Meloree » Sat Aug 29, 2009 2:13 pm

cordelia wrote:
Meloree wrote:A group of healers in a raid has nearly unlimited tools to save a tank if they know a spike is coming, or if the spike can be drawn out long enough to enable them to react.


I'm unconvinced of this. 25k melee hits landing every 2 seconds requires close to 13k HPS to heal through. Add in some dots or AoE damage, and you probably need around 16-17k HPS. I've seen lots of tank deaths because of sustained 4-6 seconds of spike damage. The initial spike was healed up, but the prolonged incoming damage out-damaged the increased incoming heals. Remember that when you assign 2 healers to MT, and the rest to the raid, when raid healing is necessary, you can only rely on the emergency buttons and incoming heals of the two MT healers. Furthermore, sometimes the oh-shit healing cd's are down, NS, penance, holy shock. That slows down max incoming HPS quite a lot.


16-17k HPS is roughly one holy paladin, plus someone else paying a little bit of attention on the side. We're in an era of near unlimited static throughput. Holy Shock isn't an oh-shit healing cooldown, it's an "oh shit, I have to move, do something" cooldown. Holy Shock, in my guild, is referred to as the tank-killer, because it creates a 2.3 second healing gap from the holy paladin who is expected to have a holy light coming every 1.2 seconds in "tank danger" kinds of scenarios. The rest of the healers are expecting a certain level of throughput, when a holy paladin uses holy shock, it's generally accompanied on vent by a comment along the lines of "pick up on MT, moving". Most of the time, that situation is predictable, or announced, and therefore covered in advance.

I'm only speaking from my experience, and from discussions with my healers. To them, a 53k "manasponge" (although, I have roughly 61% raidbuffed avoidance and over 30k raidbuffed armor, too), is significantly easier to heal in virtually all situations than a 45k tank with loads of avoidance. It may be all psychological, it may be the fact that I'm testing a lot of this out while ulduar is all on farm. But I'd consider (anecdotally, again) roughly 45k hp to be a minimum threshold for MTing Ulduar hardmodes, so, really, until you have everything on farm, it's tough to have the gear to gear any way OTHER than pure EH. The reason they like the large healthpool better is simply that they feel like they have more of a buffer in which to react, and that when they reset the spike (top up the tank), it's a longer time before the tank is in danger again. That could just be psychological. It probably is at least partially psychological. But a huge part of being an effective MT is earning the total confidence of your raid. Your DPS needs to believe that they absolutely cannot pull aggro in order to feel like they can unleash. They need to KNOW that they're safe to dps. Your healers need to believe that you're doing your best to make thier life easy, and in some ways that belief is more important than the raw numbers. Maybe it's possible to convince them of the other case, but I'm pretty unconvinved myself.

I like avoidance. In TBC, I was a huge fan, given that healers were stressed in Sunwell, and they could dial their output to the damage incoming. In a lot of ways, the combination of avoidance drs and the removal of downranking is what killed avoidance setups for me. Bandwidth is nearly irrelevant, and anti-spike tools are on short cooldowns and nearly ubiquitous. If you're starting to talk about spikes without Swiftmends and PW:S's available to help cover, you've moved into talking about sustained damage. And unless you're not bringing any spare innervates, bandwidth is just a matter of how much mana you can feed to your token holy heal turret.

But it's easy to look at the fight-specific "tank danger" situations. There are 5. Algalon, Steelbreaker, Mimiron P1, Thorim, and Yogg-0. Is there any one of those that rewards avoidance over EH in any significant way? Thorim, probably, once you can get through the threshold of Unbalancing+melee for minimum EH. But we plate tanks lose 24% avoidance, as opposed to a druid losing 13%, it's already foolish to not use a druid in progression. Everything else has either significant magical damage, diminishing the value of avoidance, or has a high enough minimum EH requirement to make avoidance stacking impractical. Ergo, for progression, like it or not, I pretend that every socket is blue, and I value an 18 stam shield enchant over a 20 defence enchant.

I'm not in the best guild in the world, but I think we do pretty well for a 16 hour a week raiding guild. I also don't ever stop trying to improve, so I'd really really love to see a convincing case made for balancing my gemming and enchants more. I haven't really seen a good model made for either case, so I'm currently really just going with intuition and experience, and it's a poor substitute for math.

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

Re: Calculating the avoidance value of expertise

Postby cordelia » Sat Aug 29, 2009 3:59 pm

We're on a major tangent here, so I'll keep this short.

You have two conditions here, both bolded by me.

Meloree wrote:A group of healers in a raid has nearly unlimited tools to save a tank if they know a spike is coming, or if the spike can be drawn out long enough to enable them to react. And any simulation that doesn't include reaction from a group of healers would generate a flawed analysis, in my opinion. Constant Stream of Healing, though, toss it out the window before you start, it doesn't match reality.


I have no problem with the former condition. It's the latter condition that I'm unconvinced of. I have seen many tank deaths 4-6 seconds after a spike begins. Isn't 4 seconds long enough for a healer to react? Their shortest spells are all <2 seconds. Isn't 6 seconds long enough for them to react? By your statement, we should never see tank deaths where the spike string lasts longer than 3 seconds. Maybe my guild just has poor healers, but we do fairly well in progression, 5/9 hardmodes, and 3rd on the server, so the evidence doesn't support that.
User avatar
cordelia
 
Posts: 630
Joined: Tue Oct 09, 2007 3:33 am

Re: Calculating the avoidance value of expertise

Postby Meloree » Sat Aug 29, 2009 5:17 pm

cordelia wrote:
Meloree wrote:A group of healers in a raid has nearly unlimited tools to save a tank if they know a spike is coming, or if the spike can be drawn out long enough to enable them to react.


I have no problem with the former condition. It's the latter condition that I'm unconvinced of. I have seen many tank deaths 4-6 seconds after a spike begins. Isn't 4 seconds long enough for a healer to react? Their shortest spells are all <2 seconds. Isn't 6 seconds long enough for them to react? By your statement, we should never see tank deaths where the spike string lasts longer than 3 seconds. Maybe my guild just has poor healers, but we do fairly well in progression, 5/9 hardmodes, and 3rd on the server, so the evidence doesn't support that.


We're probably speaking too generically to actually resolve anything. I tend to fail at communicating my point in text, but I think we may be envisioning different scenarios, because my gut reaction to your statement is "No. You should never see tank deaths to spike damage given north of 3 seconds to respond to greater than normal throughput damage." However, we can save this for another thread.

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

Re: Calculating the avoidance value of expertise

Postby knaughty » Sat Aug 29, 2009 5:59 pm

"Spike" damage lasting 6 seconds from a huge string of bad luck on avoidance (perhaps combined with parry-haste) isn't going to be solved by "other healers cover the slack".

For hard content in Ulduar, the first two things you do are have the minimum number of tanks (little flexibility) and then minimum number of healers. Every hard-mode I can think of has an in-play enrage timer - you have to stack DPS to win. Other than GV, they also tend to have a ton of raid damage.

Raid healers switching to the tank just leads to dead raid members.

And tank health spikes up and down hugely in Ulduar. How are the raid healers supposed to tell the difference between "normal" wildly swing tank health and "tank will die" wildly swinging tank health? It seems a bit "hindsight is 20:20". Tank dies, raid leader asks why, tanks and healers don't know, because everything seemed to be going fine. Healers were healing, tank was facing boss, etc. Then tank reads his log and discovers he collected 4-5 of the last 6 hits, and didn't get topped off for 6 seconds or so. If you had to consult your log to work out why you died, it's irrational to assume a raid healer can notice and switch targets.

==================

As to the original point of the thread, one question, one observation.

The question: Where does this put expertise for warrior tanks? It would have to be close to as good as dodge rating for avoidance...

The observation: My understanding of the thread is that Expertise is 70-80% as good as dodge rating for long term hit avoidance.

This makes it extremely good avoidance in practice, since the attacks you're avoiding are all burst-damage attacks from parry haste, rather than generic avoidance. I know others have said basically the same thing - but if the long-term number is 75%, I'd actually rate overall effectiveness as better than dodge rating. Bonus points: Some TPS.
This isn't the "Offtankadin" forum. My MoP FAQ: http://tinyurl.com/FAQ-5-0
- Knaughty.
User avatar
knaughty
Maintankadonor
 
Posts: 3558
Joined: Mon Dec 17, 2007 10:06 pm
Location: Sydney, plotting my next diatribe against the forces of ignorance!

Re: Calculating the avoidance value of expertise

Postby cordelia » Sat Aug 29, 2009 6:44 pm

Knaughty wrote:And tank health spikes up and down hugely in Ulduar. How are the raid healers supposed to tell the difference between "normal" wildly swing tank health and "tank will die" wildly swinging tank health? It seems a bit "hindsight is 20:20". Tank dies, raid leader asks why, tanks and healers don't know, because everything seemed to be going fine. Healers were healing, tank was facing boss, etc. Then tank reads his log and discovers he collected 4-5 of the last 6 hits, and didn't get topped off for 6 seconds or so. If you had to consult your log to work out why you died, it's irrational to assume a raid healer can notice and switch targets.


This is exactly what I see in practice. 2-4 seconds in, the tank is healed up to 70-80% health, but that still means the tank is susceptible to being 2-shot, which is what inevitably happens.

Knaghty wrote:The question: Where does this put expertise for warrior tanks? It would have to be close to as good as dodge rating for avoidance...

Do warrior tanks actually have a parryable attack faster than once every 0.53 seconds? That's what tankadin attacks work out to with a 1.3 swing speed, HotR, and SoV.

This makes it extremely good avoidance in practice, since the attacks you're avoiding are all burst-damage attacks from parry haste, rather than generic avoidance. I know others have said basically the same thing - but if the long-term number is 75%, I'd actually rate overall effectiveness as better than dodge rating. Bonus points: Some TPS.


Again, that's the discussion right above us. I think most people agree to your "gut feeling," and I have the same gut feeling, but IMO, the jury's still out. The interaction between avoidance and parry-haste in terms of tank spike damage is too complex too rely on "gut feelings."
User avatar
cordelia
 
Posts: 630
Joined: Tue Oct 09, 2007 3:33 am

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