how are you determining which ability is the most beneficial? TDR?
I wrote this neat addon that estimates Shield Barrier and Shield Block mitigation and indicates what ability will be most beneficial at any given time. The addon also displays remaining duration, remaining absorb on Shield Barrier when active, and total amount blocked by Shield Block when active.
The addon can be found here:
Here, the (min(60, rage) / 60)-part is the scaling.Code:absorb = max(2 * (attackPower - 2 * strength), stamina * 2.5) * (min(60, rage) / 60)
For Shield Block I keep track of incoming, blockable, damage during the last 6 seconds. Then I can tell how much you would have mitigated if Shield Block had been active. I also take your critical block from mastery into account when estimating Shield Block mitigation.
The addon then simply compares the two numbers (it also displays each of them) and indicates what ability is most feasible.
Sbar does scale linearly with Rage. I wasn't questioning how you calculated the exact values, i was asking what metric do you use to determine whether SB or Sbar is more appropriate.
So, you are just comparing Total Damage Reduction?
Last edited by Tengenstein; 11-05-2012 at 01:25 AM.
And yes, I simply compare total damage reduction. If you have a better suggestion I'm all ears
Does the addon differentiate between blockable and non-blockable damage or melee and magic damage? What about cases like the first boss in HoF where you have phases of blockable melee swings followed by a high physical non-blockable damage phase. How would the addon cope with this kind of situation?
Would the addon blindly trust its damage taken metric and recommend SB when the mechanics of the damage favour SBar instead?
I think the addon is potentially great, the way it presents the information is awesome, get an in game config UI and it would be stellar, but how it decides what to suggest you use just seems to be to much of a simplification. The metric is poor so the suggestion is likewise. Ideally i'd like to be able to move and resize the frames in game, and separate the Sbar and SB portions of it. Most of all just getting rid of flashy gold border when the addon thinks i should something, the option to turn that off would be really really nice.
EDIT: the only time TDR matters is if all your healers are dead or oom, in which case your raid has bigger issues than the damage the tank is taking.
Last edited by Tengenstein; 11-05-2012 at 06:34 AM.
As a side note: The non-blockable damage you speak of is also non-absorbable so you are screwed either way
Last edited by Twattle; 11-05-2012 at 07:10 AM.
Regarding TDR as a metric I don't think I totally agree with you. I agree that you want to avoid large damage spikes, but in my experience Shield Barrier is just as well suited for that job as Shield Block is, provided that it has the highest TDR. Let me clarify: First of all, I think we can agree that you shouldn't wait using Shield Barrier until you are at 60 rage, since it scales linearly anyways. Second, the point at which it begins to get "unsmooth" is when your Shield Barrier gets eaten away quickly and it takes a long time to build rage for a new one. In my experience, this is exactly the point where Shield Block begins to show the highest TDR. I can't recall taking really spiky damage while continuously popping small Shield Barriers as the addon reported it as having the highest TDR. But then again, I am only doing a bit of Mogu'shan Vaults 10-man, so if you have another experience please tell. Also, if you have an idea of how to incorporate damage spikyness in the metric, please tell.
Incorporating damage "spikyness" as a metric has always been hard, this was the focal point of the discussion back on WotLK with respect to EHP vs. Avoidance tanking. You're comparing two very different things that are relevant in two very different realms. It's like saying X class is the best DPS because they have the highest burst when in PvP you might care about that but in PvE even if you have high burst that means jack crap if you're still lowest total DPS. Kind of a bad analogy but it's sort of the inverse for tanking. If your total damage taken is low that's great, but if you take large spike damage you're still going to die because that's hard to heal against. It's far better to take more damage that is less spiky than less damage but with larger spikes.
Okay, now that we're out of the sort of "large abstract picture" let's get into the nitty gritty about shield block vs. shield barrier. And honestly, it completely depends on the boss fight and the functions of damage. For example: how much melee swings hurt. Blocks, and subsequently critical blocks, are not a raw number like shield barrier is. It's a percentage, so the harder the hit, the better it is to block something because it's higher damage reduction per rage. The other benefits come in the form of rage generation (crit blocks are more likely while sblock is up which gives you more rage) and duration.
A hard hitting melee boss can really take a toll on you if you're not mitigating the swings, and now much is mitigated, again will depend on boss mechanics and your rage generation, but let's say you've got a big sbar saved up and the boss swings at you and you absorb all of it. Now you have to hope that no damage in any other form comes in before you get your next sbar up, or you'll probably take a big hit to the face, which makes healers panic and lose mana. Healers prefer consistent damage intake with good room for error to heal you in time. But even if you get another (smaller) sbar up, will the smaller sbar absorb as much damage as the shield block? And be more consistent? The answer is no.
Now, this does often times depend on boss mechanics, sources of incoming damage, etc. For example if you are the tank tanking two of the guards on the first boss in MV, shield block is DEFINITELY going to be better because it will prevent more damage and less spikey damage. Shield Barrier is going to be GCD limited and will not make for a smooth damage profile compared to sblock, even if you can absorb more of the bleed damage which is a high percentage of your damage intake. The bleed damage is predictable and consistent though, that's not the hard part to heal. The hard part to heal is if both guards land unblocked melee attacks on you at the same time. That's when your healers panic or you have to blow a cooldown or they lose mana, and if they occurs at the wrong time, like when an explosion comes off, you can die and wipe the raid. Will you 100% of the time? No. Is it possible to just use sbar? Probably. Is it optimal? no. Is it higher TDR? Maybe, but it is much much harder to heal.
The converse example would be if you are MTing Feng. In phase 1 it's better to use Sbar because the damage is slow and predictable and you can't block while stunned. Sblock is much much better for phase 2. But for Phase 2 he hits too hard and too fast to just spam Sbar and the fire damage is predictable, again it's taking big melee hits that will wig your healers out. I found the healers preferred it when I did sblock as much as possible and just filled in with sbar as needed, especially after banking rage during a taunt rotation or something. Often times right when you taunt is the hardest time to keep a tank up because healers are still trying to heal up the previous tank while healing you at the same time. Having a full 120 rage and dropping both sblock and sbar here really helps with that.
These are just two examples, I'm sure there are plenty more.
The point here isn't that shield block is better and should always be used over sbar, or that sbar is better and there's no reason for warriors to have sblock on their bars at all, but rather that both of their place and you have to be intelligent about which you use on what boss mechanics to be the best "tank". Instantaneous/burst damage IS more important than total damage reduction as a tank, but sometimes sbar can give better instantaneous/burst reduction, depending on the boss mechanic. And sometimes sblock is up and you need to fill gaps with a trinket or sbar or other cooldowns. But overall the best "damage profile" you will get from a boss where melee swings hurt is going to be with shield block, and that doesn't even count the other benefits of blocking or crit blocking and damage scaling and yadda yadda yadda.
Anyways, TL;DR, shield block is sort of the "bread and butter" but shield barrier is definitely useful and has its place in EVERY boss fight. Both need to be used and used intelligently to have the best damage intake profile. Using one and not the other is NOT the answer to this question. Can you get by with just sbar in normals? Probably. If I were to hop on your toon and make intelligent active mitigation decisions instead of blindly pressing sbar would your healers prefer it? Probably.
Honestly there's a lot of bitching about warriors I feel like, especially wrt sblock and sbar and hit/exp/mastery, but the more and more I play more warrior, the more and more I feel like blizzard just nailed prot warriors and made it fun to tank and use active decision making for specific abilities instead of just a ridiculously simple blind man's rotation that you use sitting in front of a boss mindlessly pressing buttons while you watch TV and you could screw up the rotation miserably and no one would notice. Not so much any more, and I LIKE it.
Except for level 90 talents for prot, blizz really dropped the ball on that. WTF bliz?
Edits: I made them.
Last edited by Aggathon; 11-05-2012 at 07:52 AM.
Post 2: now what would make this addon awesome is if it could calculate damage profiles on the fly based on parsing your combat log. If you could see if you had lots of peaks and valleys and what those looked like during what abilities you were using and put it in some sort of damage taken or healing required real time graph, that would be AMAZING. I'd use the shit out of that because a lot of the "damage smoothing" and "damage profile" stuff I talk about is, admittedly, guess work where I go, "okay healers, what was better, 1 or 2? 1 or 2? okay now 2 or 3? 2 or 3? Okay so 2? Okay now 2 or 1?" And then just kinda looking at logs and trying to compare previous attempts. It's really a pain in the ass. If you had something that could compile that on a simple on-the-fly readout it would be hands down the best tanking add-on ever.
What would also be awesome is if the graph was based off of like hit points remaining and instantaneous healer mana used or something. ACTUALLY... what would be really neat would be to compare mana per second used by your healers depending on ability. Right? 'cuz the whole assumption is burst time and healer mana. If you didn't die they technically your burst time was low enough, so the next metric is stress on healers.
Damn I wish I didn't hate coding with a burning passion and love data so very very much.
Thanks a lot for your input Aggathon, although at first it sounded like you were arguing that the abilities are situational, on which we all agreed in the first place.
I think it is indeed possible to create that addon of your dreams, but personaly I don't have the time.
I still think that TDR is a pretty good metric to display as long as you know the fight and act accordingly. In the Will of the Emperor encounter I always use Shield Block after devastating combos even though the addon suggests otherwise (because you were awesome and had 0 incoming damage), but that is also because in the periods between devastating combo streaks, the TDR is a lot higher for Shield Block. My point is that I am yet to experience really spiky damage by using Shield Barrier in situations where it has higher TDR than Shield Block. If you haven't tried the addon, please give it a whirl and see what you think.
I am arguing that they are situational, but what I'm saying is the situation isn't "TDR" it's "Damage Profile". I think maybe there is a jargon dispute between "TDR" and "DR". TDR being total damage reduction being better for the entire fight, versus instantaneous damage reduction and how it continues to affect damage intake for the rest of the fight. Just because I can drop a huge sbar and block 90% of an attack doesn't mean it's a good idea to take the next one in the face when I could have blocked both of them and had a much smoother damage profile.
Reviving this again to start a discussion on how to estimate spiky damage.
@Aggathon: Your "graph" is a bit too black and white. I think the situation is more like this:
Shield Barrier: __________\/\/\/\/\/\/\/\/\/
Shield Block : vvvvvvvvvvvvvvvvvvvvvvvv
Now the question is how to detect, or rather foresee, this damage pattern. If there is no chance of gigantic hits then "_______________\/\" is usually better on healer mana than "vvvvvvvvvvvvvvvv", but where to draw the line?
I am thinking of letting the addon estimate incoming dps for each ability and chose the one with the lowest maximum over a certain time period (6 seconds for example). The downside to this is that if the Shield Barrier lasts for almost the entire duration, but only mitigates 29% of the last hit, then Shield Block will show as superior even though Shield Barrier would have been much lighter on healer mana.
What are your thoughts?
In-game UI first.
You can't forsee random game events like when you're going to take a hit string just before a big unblockable hit. I disagree that not taking any damage and then taking two massive hits is better for healer mana than not taking any damage and then suddenly taking two hit the face, that exactly what makes healing DPS a pain in the arse becuase they don't take damage and then they suddenly do and you have to react to it. the only time that isn't a problem with tanks is if the damage is trivial and if tank damage is trivial tank survival and all things related to them are a a non-issue.
How your addon represents the abilities is fine, you need to sour out an in game ui, but either use a metric like agg suggested with healing required or mana used
I never said that not taking any damage and then taking two massive hits was good for healer mana (or for yourself for that matter). Let me exemplify a bit:
Say you take 101 1k hits within 6 seconds. A 100k Shield Barrier will result in 1k dmg, the 1k being the highest hit. Shield Block would result in 70.7k dmg (not accounting for critical blocks) with 700 being the highest hit. What I mean is that obviously Shield Barrier is best in this situation, but it is hard to detect.
I guess that by in-game UI you mean a UI for configuration? Maybe I'll add some chat-commands and a lockable drag-handle, but that is all. I am better motivated to work on the metrics, since that improves my own experience with the addon.
But going back to the metrics; I don't think it is possible to predict how much healer mana is being spent on you. My best ide so far is to use som kind of weighted comparison using TDR and Highest DPS Spike (HDS), maybe 0.2 * TDR + 0.8 * HDS. The weights could be customizable of course.
right so assuming you have a 50% damage reduction from armour the raw damage is 2K, swing time ~0.06 seconds which means you're going to get 180 AP from Vengeance and your 60 rage Sbar is going to absorb 360 damage (ignoring stam scaling minmum and +10%AP). as long as ALL of the damage and vengeance is coming in the form of Blockable damage Sbar just won't be big enough to last near it's duration, you'd need a very slow swing timer (Qiang or chimaeron) or a huge RPS so that you can get 60 rage Sbars up for each and every hit. or another non-blockable source of damage that pump up vengeance, which usually would be a great place to use Sbar like on say Zor'lok's exhales or force and verve.
I still don't think you understand what I mean. Atleast your answer seems unrelated to the topic of predicting what ability is best in a given situation.
You don't need to argument for or against a certain damage scenario. I assume that all of them are possible and want to predict a damage pattern in order to find the best ability amongst Shield Barrier and Shield Block.
Last edited by Twattle; 11-20-2012 at 12:10 AM.
Not to jump in to a thread without having followed it at all...but actually yeah that's exactly what I'm going to do...
If you really want to have a good predictive model you should record damage patterns on bosses and the gear people are wearing when doing it and the abilities they're using into a big database and then run some real statistical analysis on it to give you a good predictive model. It's not going to be particularly useful to small-sample guess against this sort of thing. Plus the viability of TDR as a metric at all is questionable at best. What you want to be able to have is a model that predicts the smoothest possible damage intake for a given set of gear on a given encounter...and that requires a LOT of data to determine properly.
The problem you're trying to solve is probably a solvable one (and, honestly, a pretty interesting one), but the method you're going about it isn't apt to get you (necessarily) interesting results. Someone like WoL might be able to if they were willing to run the statistical analyses...but even they would need to fetch more data than they collect.
Now, you can probably make pretty decent guesses even in a simple mod, but it's going to be extremely tricky...
Originally Posted by Gravy