Page 11 of 17 FirstFirst ...
9
10
11
12
13
... LastLast
  1. #201
    Deleted
    I have expiremented myself a bit in mythic+ AND raids and i would say stop at 25% (23% ish + food buff to get to 25% ideally) and go for stats like mastery and versa.

    This is what i found to be the best all arounder instead of going for 30% haste.

    But then again opinions are just like assholes and everyone has one.

  2. #202
    Deleted
    Quote Originally Posted by Chrno View Post
    i am so jalous of all you people with the legendary gloves ^^ ... i really hope the stars allign at some point so that i can get my hands on those as well ;p
    You will get sephuz, like my off tank buddy warrior who got sephuz the day I got my koku's

  3. #203
    Got legendary wrists, I do nice healing :P

  4. #204
    Stood in the Fire Llarold's Avatar
    Join Date
    Apr 2011
    Location
    Altlanta, Ga
    Posts
    474
    Quote Originally Posted by Von Bosch View Post
    i would say stop at 25% (23% ish + food buff to get to 25% ideally) and go for stats like mastery and versa.

    This is what i found to be the best all arounder instead of going for 30% haste.
    Shooting past 25% Haste is largely about reducing randomness. Devastate can be pretty fickle, and just not proc Shield Slam for 8+ hits in a row. Those spots can get rough, and that extra 3-5% Haste can make a lot of difference in keeping your health stable in the worst-case scenarios. When luck is on your side, stopping at 25% Haste and pushing Mastery is definitely better.

  5. #205
    TL;DR Per point Haste (100% effective uptime) > Mastery > Versatility > Strength / Critical Strike > Critical Strike / Strength > Haste (after 100% effective uptime)

    Roughly 30% haste is needed for close enough to 100% uptime; its value is reduced when you are not always tanking.

    This does not mean you 'need' a certain amount of haste. If an item is a notably higher item level, it's probably better. Using a better trinket is good, even if you 'lose haste'.

    Defensive Stats for Protection Warrior in Legion

    Written by me (Macrologia), with much assistance from Marok, and contributions from gray_hound.

    Abstract

    Unfortunately, SimCraft is currently not working properly for many reasons. Ask Mr Robot’s sims produce metrics which are not conducive to finding any kind of stat priority for a meaningful purpose. As such, we cannot provide ‘stat weights’.

    However, we can prove that haste is better than mastery; that mastery is better than versatility; and that versatility is better than strength. Crit is obviously worse than the other secondaries, but we don’t know exactly where it is relative to strength, or how much worse it is (realistically it’s probably not that much worse).

    We will assume that all damage is blockable, because the vast majority of relevant damage in raids and dungeons is blockable. If you wanted a second gear set designed specifically to counter magic damage, that gear set should be prioritising versatility above all else; but if, like most, you only use one gear set, it should be designed at reducing physical, blockable damage as much as possible.

    Haste vs. Mastery

    Due to haste reducing the cooldown of Shield Block, but mostly due to the talent Heavy Repercussions, haste allows for a much larger uptime on Shield Block. Though Simcraft is broken and Ask Mr Robot produces results in poor metrics, Ask Mr Robot’s sims were working sufficiently for the purpose of determining how much haste contributes to Shield Block uptime.

    The APL used is as follows:

    actions.prot=shield_block,if=!buff.neltharions_fury.up&(cooldown.shield_slam.remains=0|act ion.shield_block.charges=2)
    actions.prot+=/neltharions_fury,if=!buff.shield_block.up&(cooldown.shield_block.remains>3&action.shield_b lock.charges=0)&cooldown.shield_slam.remains>3
    actions.prot+=/shield_slam,if=!(cooldown.shield_block.remains<=gcd.max*2&!buff.shield_block.up)
    actions.prot+=/revenge,if=cooldown.shield_slam.remains<=gcd.max*1.5|cooldown.shield_block.remains<=gcd.ma x*2&!buff.shield_block.up
    actions.prot+=/devastate

    (Note that it is coded such that you generate 10 rage via intercepting automatically on cooldown)

    This APL is given in SimC syntax to make it easier to read for those familiar with SimC, but the sims were performed in AMR.

    This uses shield block, as long as neltharion’s fury is not up, and shield slam is ready; or if neltharion’s fury is not up, and shield block has 2 charges.
    Then it uses neltharion’s fury, as long as shield block is not up, and the cooldown on shield block is longer than the duration of neltharion’s fury, and the cooldown on shield slam is longer than the duration of neltharion’s fury.
    Then it prioritises shield slam, but won’t use it if shield block is not up, and about to come off cooldown. This will only cause it to delay shield slam much if you’re on low haste, but is optimal for shield block uptime either way.
    It only uses revenge in the global cooldown preceding shield slam coming off cooldown on its own.
    Lastly, if nothing else is available, it uses devastate.

    This APL makes some assumptions. Firstly, it assumes that you will always have enough rage to cast shield block. This is very likely to be the case. Secondly, it assumes that you don’t need to cast shield block at inopportune times for the purpose of surviving the boss’ hits. This is not likely to always be the case in practice, though it will matter less as you get more haste.

    To calculate damage reductions from haste and mastery put together, we can use the following.

    With shield block up, you get a 30% damage reduction, plus a chance for it to be a 60% reduction instead based on your mastery. When shield block is not up, you get no damage reduction, plus a chance to reduce the damage by 30% based on your mastery, plus a chance to reduce it by 60% instead based on your mastery.

    At 0 mastery, you have a 4% chance to block, and a 12% chance for your blocks to be critical blocks, baseline. Per point, mastery gives an additional roughly (29/6812)% chance for a block to be a critical block.

    The formula for calculating block chance (which has diminishing returns) is, as a percentage:
    ((MasteryRating / 350 + 8) * 0.5) / (((MasteryRating / 350 + 8) * 0.5) * 0.018 + (1 / 0.94)) + 3 + 15

    Accounting for shield block uptime, the formula is therefore:

    Total Damage Reduction = 1 – (Uptime*critblockchance*0.4 + uptime*0.7*(1-critblockchance) + (1-Uptime)((1-blockchance)*1 + blockchance*critblockchance*0.4 + blockchance*(1-critblockchance)*0.7))

    Though SimCraft is broken in numerous ways and AskMrRobot puts out weird metric results, AskMrRobot’s sims for haste and shield block uptime work properly.

    Haste Uptime
    0 0.703
    3000 0.783
    4500 0.816
    6000 0.855
    9000 0.926

    This equates to (data here):

    Note - corrected
    (Thanks @Alaexaender)



    So mastery is better than haste per point at reducing damage taken on average. However, it is nevertheless advisable to go for haste for defensive purposes anyway. This is because though mastery may reduce on average slightly more damage, haste is vastly more consistent. Having a 100% chance to block, with that block being guaranteed to be 30%, with a 12% chance for it to be 60% instead, is far more valuable than having a 30% chance to block for 30%, with a 50% chance for it to be 60% instead. It is similar to parry in that whilst 350 crit might be 1% parry, and thus 1% damage reduction on average (before diminishing returns), it is not at all reliable and much less valuable as a defensive stat in general for that reason.

    Obviously you do get fairly high shield block uptime with low haste, but it’s still worth going for haste until you get 100% uptime whilst actively tanking, so that all blockable hits are guaranteed to be blocked. When tanking all the time, this is somewhere around 30% haste; slightly higher. It is generally not advisable to go for quite as much as 100% uptime because you aren’t tanking literally 100% of the time. In raids, roughly 20% haste is more than sufficient for 100% effective uptime on the majority of raid bosses. However, I would recommend 30% haste as it’s much better in dungeons; better for damage done; and better on several raid bosses.

    This doesn’t account for mastery’s contribution to ignore pain, but the higher the incoming damage, the less relevant that is – when you’re taking threatening damage, it will likely make extremely little difference. Even if it would put mastery ahead of haste (such as when taking very low damage – not that these situations are important), haste provides vastly smoother damage taken, and more DPS.


    Mastery vs. Versatility

    The formula for Ignore Pain is roughly 18.6*AP*vers. This means that a 1% increase to mastery and a 1% increase to vers both increase Ignore Pain by the same amount, because 18.6*(AP*1.01)*vers = 18.6*AP*(vers*1.01).

    However, this means multiplying your total e.g. vers by 1.01; this is not the same thing as going from e.g. 5% vers to 6% vers. That would be multiplying your vers by 1.2.

    Going from x% mastery to (x+1)% mastery is the same as going from x% vers to (x+1%) vers, for increasing the size of Ignore Pain. However, it must be remembered that you have a base amount of mastery. Warriors have 8% AP gain from mastery baseline.

    350 mastery is required for 1% AP from mastery. 400 vers is required for 1% extra healing done by Ignore Pain.

    If you had 3200 vers and 0 mastery, then an additional 1% mastery would be equal to an additional 1% vers. This means that at 3200 vers and 0 mastery, 350 mastery = 400 vers (for increasing the size of Ignore Pain). This means that at 3200 vers and 0 mastery, mastery is better per point at increasing the size of Ignore Pain.

    At 0 vers and 0 mastery, 400 vers is a 1% increase. To get the same amount from mastery, however, you need more – you need 1.08%. This is 378 mastery. Therefore at 0 vers and 0 mastery, mastery is still better per point at increasing the size of Ignore Pain. However, if you had a lot of mastery, and very little vers, then vers would be better per point at increasing the size of Ignore Pain.

    The main function of mastery, however, is not to increase the size of Ignore Pain; it’s to reduce your damage taken by giving you critical block chance. You need about 233 mastery for 1% chance to critically block instead of blocking.

    Since you are already blocking all blockable attacks, thanks to Shield Block, then a 1% chance to critically block means a 1% chance to get a 43% damage reduction (rounded), i.e. going from taking 70% of the damage (a regular block) to 40% of the damage (a critical block).

    This means the amount of rating you need to get on average a 1% damage reduction from mastery is about 541. This does not account for Neltharion’s Fury, during which you have a guaranteed critical block chance, so mastery is worthless for this particular purpose. Neltharion’s Fury has a maximum uptime of 3/45. Though in practice it would never be used exactly on cooldown, if we reduce the effectiveness of mastery by 3/45, we see that you need about 580 mastery rating for an average of 1% damage reduction.

    Versatility is much worse than this for its damage reduction component, requiring 800 stats for 1% damage reduction.

    These results suggest that at low haste levels, versatility may be better than mastery, because you do not get as much chance to critically block via mastery if you aren’t already guaranteed to be blocking. However, this is off-set by the fact that mastery also increases your base chance to block (this simply goes to waste when shield block is up).

    As such, mastery is clearly better than versatility for reducing blockable damage taken.

    Since the vast majority of relevant damage taken is blockable, you ought therefore value mastery over versatility.

    Versatility vs. Strength

    The value of versatility relative to the value of strength depends on three things: The amount of strength you have; the amount of versatility you have; and the amount of damage you are taking before mitigation. The more strength you have, the more versatility increases Ignore Pain by. The more versatility you have, the less damage you take after mitigation. The more damage you take before mitigation, the more damage you take after mitigation.

    To calculate the value of 1 str relative to 1 vers:

    Where ‘str’ is ‘amount of strength’, and ‘vers’ is ‘multiplying factor for versatility’

    Formula for 60 rage Ignore Pain with Indomitable: 18.6*1.25*str*vers

    Value of str per point per ignore pain = 18.6*1.25*vers

    Value of vers per 1% vers per ignore pain = 18.6*1.25*str*(1/100)
    Value of vers per point per ignore pain = 18.6*1.25*str*(1/100)*(1/400)

    Value of str per ignore pain – value of vers per ignore pain = (18.6*1.25*vers) - (18.6*1.25*str*(1/100)*(1/400))

    At 1 ignore pain per roughly 10 seconds from logs:

    Value of str per second – value of vers per second = ((18.6*1.25*vers) - (18.6*1.25*str*(1/100)*(1/400)))/10

    Amount of DTPS mitigated by 1 point of vers = DTPS*0.01/800

    Where the damage mitigated by vers is greater than the difference between the amount of Ignore Pain generated by 1 point of str and the amount generated by 1 point of vers, vers is better than str.

    Alternatively put, if the following expression is true, vers is better than str:

    DTPS*0.01/800 > ((18.6*1.25*vers) - (18.6*1.25*str*(1/100)*(1/400)))/10

    Or

    DTPS > (((18.6*1.25*vers) - (18.6*1.25*str*(1/100)*(1/400)))/10)/(0.01/800)

    This simplifies to:

    DTPS > 8000(23.25*vers – 0.00058125*str)

    So if that is true, with your str and vers levels, then vers is more valuable than str for you.

    For example, at 25k str and 10% vers, vers is better than str when you are taking 88.35k DTPS. This is a very low number.

    At 25k str and 20% vers, vers is better than str when you are taking 106.95k DTPS. This is slightly higher, but still very low.

    As such, vers will always be better than str per point in content that might be even remotely dangerous.

    Str vs. Crit

    Both of these stats are not very good (but that doesn’t mean they’re awful – just not very good). Working out exactly where they are relative to the other stats is quite difficult without working sims.

    Stamina

    Stamina is a good stat for survivability – that is, for helping you to survive, or helping you avoid a risk of dying. With Indomitable, a stamina flask (1950 stamina) gives me 240975 health. In any given short period from full health, this is the same thing as reducing my damage taken by 240975 – except for the fact that it has to be healed back up afterwards. This is usually not an issue, however, and moreover, not dying in the first place is more important than reducing your healing required (though the latter is obviously important too). From figures calculated earlier, mastery gives an average of 1% damage reduction per 580 mastery (not counting the amount contributed to Ignore Pain, but this doesn’t change a huge amount here). 1950 mastery therefore gives, not counting Ignore Pain, about 3.36% damage reduction. This means that, not counting Ignore Pain, for mastery to ‘mitigate’ the same amount as stamina, in any short window, you would need to be taking over 7 million damage in that short window. This is quite a high number. Obviously Ignore Pain is relevant, and so is the fact that damage ‘mitigated’ by Stamina needs to be healed back up whereas damage properly mitigated doesn’t. But this doesn’t change the fact that Stamina is good at helping you avoid the risk of dying.

    Conclusions

    Haste is the best, increasing shield block uptime.
    Mastery is the next best, increasing your chance for blocks to be critical blocks; increasing your chance to block when shield block is not up; and increasing the size of Ignore Pain.
    Versatility comes after, increasing the size of Ignore Pain and reducing your damage taken.
    Strength and Crit are after Haste, Mastery and Versatility. Strength increases the size of Ignore Pain, and gives a small amount of Parry; Crit gives a larger amount of Parry.

    Therefore, looking at each stat's contribution to reducing total damage taken, DPS, and smoothness, the recommended stat priority is as follows:

    Haste (100% effective uptime) > Mastery > Versatility > Strength / Critical Strike > Critical Strike / Strength > Haste (after 100% effective uptime)

    Roughly 30% haste is needed for close enough to 100% uptime; its value is reduced when you are not always tanking.

    This does not mean you 'need' a certain amount of haste. If an item is a notably higher item level, it's probably better. Using a better trinket is good, even if you 'lose haste'.
    Last edited by Lysozyme; 2016-11-11 at 01:23 PM.

  6. #206
    Very insightful post, thank you all for the work.

    2 questions remain for me. I know that no definite answer is possible in any way, so I wonder how you would do it:

    1. Now without working sims most people would probably hope for a guesstimate, how many itemlevel (if any at all) you would sacrifice for a more favorable stat combination?

    2. How would you rate the stats (including strength) for dps purposes? Probably something like Haste (30%) > Critical Strike > Haste (after 30%) > Mastery > Versatility? Where would Strength fit in there? Again, even guesstimates would be nice to see

    Thanks

  7. #207
    Mastery and vers are probably both better than haste after 30% for dps

    Your other question - it really depends on a lot of things
    Macrologia (expert on Protection Warrior and Guardian Druid)

    Skyhold

    Dreamgrove

  8. #208
    Pandaren Monk Chrno's Avatar
    Join Date
    Jun 2010
    Location
    Westland
    Posts
    1,864
    Quote Originally Posted by daed4 View Post
    You will get sephuz, like my off tank buddy warrior who got sephuz the day I got my koku's
    tbh ... i think that ring is amazing for mythic +
    i'd love to get it some day to experiment with it ^^

    Shockwave + fearing trash would be a hell of fun :P
    Warrior, getting my face smashed in because I love it

    "The Perfect Raid Design Drawn by me .

  9. #209
    Very interesting post Lysozyme, and very much appreciated. If 30% is the magic number for haste, what might
    be the absolute minimum amount of Mastery you should have to offset all that haste? or is there no minimum amount?

  10. #210
    No minimum amount of mastery. Maximum would be 100% chance to critically block which is about 24k mastery (realistically unreachable).

    30% haste isn't actually a hard cap - you get 100% uptime at something like 37% haste on average, though obviously if you're slightly above average with rng on a particular attempt, a lot of that goes to waste - but 30% haste will be enough to get 100% effective uptime in virtually any situation
    Macrologia (expert on Protection Warrior and Guardian Druid)

    Skyhold

    Dreamgrove

  11. #211
    Ill give this a try. At 25% haste im at 50% mastery and 3% verse. Not sure how bad that 3% verse is.
    Normally I tend to favor a 16% H 57% M 7% V build, that just feels right.

  12. #212
    Thanks for all your work Lysozyme, Marok, and gray_hound. It is much appreciated. I drew basically the same conclusions about Mastery and Vers using some amateurish spreadsheet-crafting a while back, but I wasn't sure if I was right so I wasn't sharing anything publicly. Great to get a more sophisticated analysis. And I had no idea about Strength. Super helpful. We've been in disarray over stats for far too long.

    One question: When I use your Str vs Vers metric, I get different numbers than your post has. E.g., you say that with 25k Str and 10% Vers, DTPS would need to be >88.35k to make Vers more valuable than Str. But when I plug those number into your formula -- (8000*((23.25*vers)-(0.00058125*str)) -- I get a negative number(!), the absolute value of which is still different: 97.7k. Using 25k Str and 20% Vers, I get -79.1k.

    I didn't go back and work through all your formulas and simplifications. Maybe there's just a typo in the final formulation? Or maybe I'm missing something.

    Thanks again.

    - - - Updated - - -

    One more thing I've been wondering about: What's with AMR valuing Leech so highly? How should we value Leech? Same with Avoidance (though AMR doesn't value it nearly as much as Leech).
    Last edited by Brewschii; 2016-11-03 at 12:11 PM.

  13. #213
    Quote Originally Posted by Lysozyme View Post
    Since the vast majority of relevant damage taken is blockable, you ought therefore value mastery over versatility.
    This is a HUGE assumption that you make for literally all of your calculations. And it's an offhand comment buried in the middle of your post. It's also blatantly false on every single boss fight in Emerald Nightmare. At best, you're looking at physical, blockable damage accounting for 60% of your total damage taken on a fight. More realistically, blockable damage is 40-50% of a current boss fight in Emerald Nightmare.

    Drake tank on Cenarius 55% blockable damage
    Boss tank on Cenarius 60% blockable damage
    Xavius tank 36% blockable damage
    Ursoc tank 30-40% blockable damage

    Those are considered the benchmark fights in Emerald Nightmare currently and none of them have "a vast majority of blockable damage." Nor, is the blockable damage the most threatening.

    Haste and Mastery look like god-tier stats when you pretend that 100% of the incoming damage is affected by them fully. If you take the values you're getting for Haste and multiply it by .5-.6 and then take Mastery and multiply it by .65-.75 (due to it still affecting IP) you'll get much more realistic numbers. And that's where Versatility starts to make an impact.

    Your post also pretends that you need anywhere close to 100% shield block uptime on boss fights. Again, this is completely not true.

    68% shield block uptime on Xavius
    96% blocked hits (same log, mouse over melee block to see percent blocked)

    86% Shield Block uptime on Cenarius
    94% blocked hits (same log)

    And Clutch could have just played better. With ~80% shield block uptime he would have 100% blocked melee hits. This illusion that anywhere close to 100% shield block uptime is a good thing just needs to freaking stop (it's slightly better in 5 man content as you mention).
    Last edited by Emancptr; 2016-11-01 at 03:55 PM.

  14. #214
    Quote Originally Posted by Emancptr View Post
    This is a HUGE assumption that you make for literally all of your calculations. And it's an offhand comment buried in the middle of your post. It's also blatantly false on every single boss fight in Emerald Nightmare. At best, you're looking at physical, blockable damage accounting for 60% of your total damage taken on a fight. More realistically, blockable damage is 40-50% of a current boss fight in Emerald Nightmare.

    Drake tank on Cenarius 55% blockable damage
    Boss tank on Cenarius 60% blockable damage
    Xavius tank 36% blockable damage
    Ursoc tank 30-40% blockable damage

    Those are considered the benchmark fights in Emerald Nightmare currently and none of them have "a vast majority of blockable damage." Nor, is the blockable damage the most threatening.

    Haste and Mastery look like god-tier stats when you pretend that 100% of the incoming damage is affected by them fully. If you take the values you're getting for Haste and multiply it by .5-.6 and then take Mastery and multiply it by .65-.75 (due to it still affecting IP) you'll get much more realistic numbers. And that's where Versatility starts to make an impact.

    Your post also pretends that you need anywhere close to 100% shield block uptime on boss fights. Again, this is completely not true.

    68% shield block uptime on Xavius
    96% blocked hits (same log, mouse over melee block to see percent blocked)

    86% Shield Block uptime on Cenarius
    94% blocked hits (same log)

    And Clutch could have just played better. With ~80% shield block uptime he would have 100% blocked melee hits. This illusion that anywhere close to 100% shield block uptime is a good thing just needs to freaking stop (it's slightly better in 5 man content as you mention).
    Your damage totals from logs ignore damage that was blocked. Take your link for "Boss tank on Cenarius". The log says 72.18m melee, 118.68m total damage. From that, you get 60% melee damage. But 72.18m is the figure after blocks. The actual figure for incoming melee damage (before blocks) was 126.8m, making the real figure for total damage 173.4m. So blockable damage was 73%. I only did the math on this one log, not the others.

    That said, your general point about the value of Vers relative to Mastery (set Haste aside for now) is not wrong. It's definitely true that at a certain point of magic damage vs melee damage, the value of Vers can exceed Mastery. Using my spreadsheet and some logs, I thought I figured out that physical damage outweighed magic enough that we could safely assume Mastery was more valuable than Vers almost all of the time (assuming adequate uptime on SB) -- and that the exceptions to this rule would be few enough, and minor enough, that we could safely ignore them in making gear choices.

    Reflecting on it now, I bet I was looking only at Heroic logs then because I didn't have any / many Mythic logs yet. Your post makes me realize there might be a heck of a lot more magic damage, proportionally, in Mythic. Which would make sense -- both because of new adds / mechanics, which are often magical, and because of the increased likelihood that tanks will fail to avoid magic damage (or that other raid members will fail mechanics that, in turn, cause the tanks to take more magic damage). There's still one counterpoint to that though: You're much more likely to use CDs (Demo Shout, Shield Wall, Spell Reflect, externals) on major magic damage. So you'd need to account for that in thinking about the overall balance of damage for gearing purposes.

    Getting the right answers to these questions probably matters, and it's certainly worth pushing on a big assumption like 100% blockable damage to figure out how much it matters.
    Last edited by Brewschii; 2016-11-01 at 05:52 PM.

  15. #215
    Quote Originally Posted by Brewschii View Post
    Your damage totals from logs ignore damage that was blocked. Take your link for "Boss tank on Cenarius". The log says 72.18m melee, 118.68m total damage. From that, you get 60% melee damage. But 72.18m is the figure after blocks. The actual figure for incoming melee damage (before blocks) was 126.8m, making the real figure for total damage 173.4m. So blockable damage was 73%. I only did the math on this one log, not the others.
    You're assuming that I block nothing if i'm not going for a Haste/Mastery build, which simply isn't true. That's the entire point that i'm making when I talk about about effective shield block uptime.

    You get ~66% Shield Block uptime (buff in warcraft logs) while using Heavy Repercussion talent and 0% haste. 66% shield block uptime is anywhere from 80-100% effective shield block uptime (% of boss melee hits blocked) on boss fights in Emerald Nightmare. The Mythic Xavius logs I linked above show exactly that (68% uptime and 96% effective). If it's required, I can go through and link more logs showing this effect. In general, effective shield block uptime will be 20-30% higher than pure shield block uptime if you're playing well.

    What that means is pushing to 30% haste has 0 effect on some boss fights (Nythendra, Cenarius depending on what you tank, Ursoc, and Xavius). This is because you already have very close to 100% effective shield block uptime.

    Edit* 0 effect is obviously an exaggeration. Haste has some effect, but it's much less than is being advertised.
    Last edited by Emancptr; 2016-11-01 at 06:59 PM.

  16. #216
    Quote Originally Posted by Emancptr View Post
    You're assuming that I block nothing if i'm not going for a Haste/Mastery build, which simply isn't true. That's the entire point that i'm making when I talk about about effective shield block uptime.
    I wasn't making any assumptions. I was just saying that, when your post talks about the percentage of damage that is blockable on various fights, it calculates that number incorrectly. If you're trying to figure out how important it is to mitigate blockable vs non-blockable damage, then you ought to use the proper amount of each one as the basis for your analysis. That's all. This point has nothing to do with whether you actually blocked the damage or not.

  17. #217
    Your number isn't correct either because it assumes that I block nothing. If I have literally 0 stats I would still not take 126.8 million damage because I have the ~66% Shield Block uptime and ~80%+ effective shield block uptime. When evaluating the value of Haste you have to look at it in terms of how many more melee swings I would block with that haste.

    My whole point is that this number is actually very small because of the relatively high effective shield block uptime baseline. My argument with the logs linked above is that those Shield Block uptimes are reachable without haste. Therefore, they are accurate representations of what % physical damage is before considering any stats.

  18. #218
    Quote Originally Posted by Emancptr View Post
    Your number isn't correct either because it assumes that I block nothing. If I have literally 0 stats I would still not take 126.8 million damage because I have the ~66% Shield Block uptime and ~80%+ effective shield block uptime. When evaluating the value of Haste you have to look at it in terms of how many more melee swings I would block with that haste.

    My whole point is that this number is actually very small because of the relatively high effective shield block uptime baseline. My argument with the logs linked above is that those Shield Block uptimes are reachable without haste. Therefore, they are accurate representations of what % physical damage is before considering any stats.
    We're talking past each other, and I've tried to clarify what I'm saying once already. I'm going to try to avoid getting sucked into an extended back and forth about getting the baseline numbers right for incoming damage. For what it's worth, I completely understand what you're saying about SB uptime versus % of blocked melees, and always have. From memory, I think have a Nythendra log where I have around 47% SB uptime but blocked 89% of hits. So I get it. I'm not talking about uptime vs blocks.

    On a separate note, you say that your logs are are examples of "literally 0 stats" and SB uptime "without haste." But in the Cenarius log, the prot war has 32.9% Haste (10,691 rating). Huh? https://www.warcraftlogs.com/reports...t=14&source=17

  19. #219
    Quote Originally Posted by Brewschii View Post
    On a separate note, you say that your logs are are examples of "literally 0 stats" and SB uptime "without haste." But in the Cenarius log, the prot war has 32.9% Haste (10,691 rating). Huh? https://www.warcraftlogs.com/reports...t=14&source=17
    Clutch is a bad example, because I know he gears for Haste. But he's the only "top end" warrior that I know and am comfortable linking his logs. Having said that lets examine those logs.

    Melee Damage Taken from Cenarius

    See those huge windows where he isn't taking melee damage (Note: it's windows he's tanking the Tree/Sisters on. But 85% of total melee damage Taken is from Cenarius. You don't need to actively Shield Block while tanking the tree/sisters). I could go through and look at the timestamps for the exact numbers, but I'm just going to rough it off the graph. He has ~2:30 where he isn't actively tanking something that requires shield block. That means given his fight length (7:07) he would need 64% Shield Block uptime in order to have 100% effective Shield Block uptime while tanking Cenarius. This means he could even Shield Block for the Tree that he tanks. All of that is the case if he had exactly 0% haste.

    Melee Damage Taken for their other Warrior tank

    The tank that has the boss second needs even lower Shield Block uptime. This tank could get away with ~50% Shield Block uptime and have 100% effective uptime.

    You are correct that I did not link a log to a tank that is not gearing for Haste and still has the uptimes that I'm talking about. However, you can simply look at those logs and see that the uptimes I'm talking about are sufficient. It only requires optimal usage of Shield Block.
    Last edited by Emancptr; 2016-11-01 at 09:37 PM.

  20. #220
    Hey -- I designed the NPS metric and tank simulator on AMR, thought I'd add a few comments of interest:

    1. We're going to give a lot more details on NPS when we release the Mythic+ tanking scripts soon-ish, and add in a Xavius script as well (to test heavier magic damage vs. physical).


    2. Regarding NPS giving "weird results" -- at least for the purposes of the Shield Block analysis that @lysozome was doing, it gives exactly the data that you want. Here's a quick example, just took a generic 865-ish warrior, set stats to about 30% haste and high mastery too:

    http://www.askmrrobot.com/wow/simula...22079e2ff01c65

    In that report is a line item for "Base Block", and grouped under it are all sources of block separated out. The colored bar if you mouse over is the total NPS from each source of block, and the "Average" column is the average per hit blocked. So here's how you interpret that and translate it into something "real":

    The "Total" value if you mouse over the dotted bar by "Base Block" is the total amount of raw damage that was blocked by any source of block (in this case, 11.5m). It's really the total "NPS", but for blocking, NPS and total damage blocked are essentially equivalent. This is generally true for any kind of damage reduction or absorb.

    Now if you total up all the "Average" values for each source of block (9k + 200k + 46k + 32k + 263k), that's the total average amount of damage blocked per hit, or about 550k here.

    The individual lines for blocking show you relatively how much each source of block is contributing -- the code is smart and deals with the "overflow" of Shield Block appropriately. The idea here is that your base block, mastery, bastion of defense, etc. all add some block, so it proportionally attributes the value of any block to the buffs that are giving block chance at the time you get hit, based on how much block chance they are giving you at that moment in time.

    Whether you "like" NPS or not, this is some really useful data that has previously been unavailable to tanks to make decisions about the relative strength of abilities, gear, etc.

    The less "obvious" aspects of NPS are to make it easier to directly relate fundamentally different kinds of toughness in a simple, quantifiable manner (e.g. how do you compare stamina to block? not at all obvious how to make that meaningful). But we don't need to get into that right now, not relevant to the current discussion.


    3. Regarding this discussion of how much damage is "blockable" in a typical fight, the answer is "a lot". Mythic+ is very melee-heavy. Most raid bosses are dominated by melee damage, 70% or so is a good "average" for a generic fight.


    4. Given the above, let's talk about blocking specifically. The report on one of our simulations shows very clearly that (depending on gear) Ignore Pain will prevent a minimum of 2x as much damage as Shield Block, and as much as 3x. You can ignore any "nuances" of NPS for this comparison, as stated above, the NPS for Ignore Pain and Shield Block is the same as total damage prevented by each ability.

    I think that a lot of people reading these discussions fail to factor in that the amount blocked is calculated AFTER damage reductions from e.g. Armor and Versatility. If you take a raw 1.15m melee hit (about the raw value for Ursoc Heroic last I tested), a critical block will not reduce the hit by 0.6 * 1.15m = 690k. Your armor will reduce the raw hit to roughly 60% of the raw value, or 690k, then you crit block 60% of THAT, or about 414k. Any other damage reductions from e.g. versatility, cooldowns, etc. will further reduce the amount that a block will prevent, so there is an interaction there that gives me a headache to try and math out... thus a simulator! Just tell the simulator to do it, and analyze the results.

    I've also seen another big misconception flying around prot warrior discussions: most people think the main value of versatility is to make Ignore Pain bigger. That is important, yes, but people seem to ignore what the stat says that it does... reduces your damage taken! Take this example simulation with a mastery/vers-heavy build, around 6700 versatility:

    http://www.askmrrobot.com/wow/simula...f4e5377a7489ea

    Versatility's passive damage reduction prevents almost 2x as much as Shield Block. If you compare to the previous haste/mastery setup, the change to Ignore Pain is not that big. It's the stat's damage reduction itself that is the most valuable part. If you just decide hey, let's ignore the best part of versatility... sure, it's going to seem like a bad stat.

    Once again, you can see the power of the NPS report in the simulator: a very straightforward way to quantify the difference between disparate sources of toughness.


    Unless someone sees something really wrong with the simulation... it seems completely reasonable to not put so much emphasis on Shield Block uptime. You can get pretty good coverage of time spent tanking with low haste. Something that makes Ignore Pain bigger is probably going to be a bigger "bang" for your gear budget in a raid tanking setting. Thus I stand by our current recommendation for an Ursoc-style fight (vers/str/mastery), and can't really recommend this haste-centered build. It's not a terrible build... no stat is so awful that you can't tank with it. But it's probably not the best choice in this case. Now it's possible that a different boss won't rank Versatility as high (because there's a significant amount of damage from e.g. Rend Flesh, a bleed), but it will probably always be a strong stat.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •