Discussions taking place on other forums [Protection]

[Dreador]

Posting only for informational purposes, I don't even think I posted in any of these threads but I wanted to share what I'm reading on other sites.

WoW Forums:
http://us.battle.net/wow/en/forum/topic/6490010693
http://us.battle.net/wow/en/forum/topic/6522913654

Theck's Blog(linked to a single post but you can expand):
http://sacredduty.net/2012/09/10/l90...-for-warriors/

Icy-Veins (guide, their forums are crickets):
http://www.icy-veins.com/protection-...rity-reforging
I definitely would continue to stay away from AskMrRobot as was smart in Cata as well.

MMOC:
http://www.mmo-champion.com/threads/...rior-Guide-5-0

ElitistJerks:
**crickets** RIP

Noxxic:
http://www.noxxic.com/wow/pve/warrior/protection
This one was not shabby, but a bit light. I didn't get the sense that I was being educated by the read so much as just told to shut up and do it. I only take that advice from the wifey, so best be backin' off, Noxxic.

Joystiq:
http://wow.joystiq.com/category/warrior/
Pretty sure this guy's head is in a place it should not be elsewhere on his body.

Again, not posting these to sway opinions or even to generate traffic to any of these sites, but I did want to see what others are talking about outside our garden.

[Tengenstein]

Pity you don't include Theck's Blog, almost all the maths for warriors seems to come from him.

[Dreador]

Pity you don't include Theck's Blog, almost all the maths for warriors seems to come from him.

You had already covered it quite well in the guide so I kinda didn't focus on it. /added

[Quietsch]

The blogpost from Theck is _terribly_ wrong.
(Nearly all) Many of the conclusions are either wrong, as they are based on wrong assumptions.

(Just mentioning it because... well, at least Theck is most of the times trustworthy... :>)

[kopcap]

Elaborate pls?

[Quietsch]

(not exhaustive) :
- wrong statweigths (which isn't suprising, as the base for this is wrong)
- wrong mastery -> cblock conversion
- wrong SnB-chance
(- meta ignored)
- sbarr totally ignored in statvalues
- I'm pretty sure, there's something wrong with his 'ability use', as those values for hit/expt are simply nonsense. (But I'm on the other hand not sure, how much of those low values is only because of the wrong mastery conversion and thus a less valuable RPS/SB-Uptime, resulting in a lower value for hit/expt as the gained SB-Uptime is... well, less valuable :>)
- possibly more (It would really help, if I was able to fully understand matlab)

[jere]

Have you posted your concerns on his blog? He's addressed everyone's points so far and backed it up with both data and math. If he still has issues, it would be good for you to alert him to it.

[Quietsch]

Someone i've discussed this with had (or at least he said it) already sent him a pm regarding this.

[Sharku]

Elitist Jerks R.I.P.? No more theorycrafting for them, guild closure or just no more cookie cutter specs for them to write guides about?

[Dreador]

Elitist Jerks R.I.P.? No more theorycrafting for them, guild closure or just no more cookie cutter specs for them to write guides about?

I have no idea what's going on, but to see what I mean just pop over to their warrior forum.

[Tengenstein]

EJ's never been particualrily great for protection.

[theckhd]

(not exhaustive) :
- wrong statweigths (which isn't suprising, as the base for this is wrong)
- wrong mastery -> cblock conversion
- wrong SnB-chance
(- meta ignored)
- sbarr totally ignored in statvalues
- I'm pretty sure, there's something wrong with his 'ability use', as those values for hit/expt are simply nonsense. (But I'm on the other hand not sure, how much of those low values is only because of the wrong mastery conversion and thus a less valuable RPS/SB-Uptime, resulting in a lower value for hit/expt as the gained SB-Uptime is... well, less valuable :>)
- possibly more (It would really help, if I was able to fully understand matlab)

A few of those have already been pointed out to me. The mastery -> cblock conversion was indeed incorrect; I'm running updated simulations with that adjustment corrected as we speak.

I hadn't noticed the SnB chance being in error, but you're right. I was using 20%, it's 30%. That's fixed in the new version as well.

Ignoring the meta (by which I assume you mean meta gem?) is basically irrelevant. The armor meta is likely to be much stronger for all tanks anyhow, and a 1% change in block value isn't going to massively shift anything.

Sbarr is ignored for good reason - it's not in the model. The idea was to adapt a simulation I already had (for paladins and SotR) and see what the corollary was for warriors. Namely, if you're trying to maximize Shield Block uptime. That's working under a number of assumptions, first of all that Shield Block is actually your go-to mitigation tool. I've read that due to balancing tweaks, Shield Barrier might actually be better at both 85 and 90. That would require a different model.

In short, the choice of SBlock over SBarr isn't inherently "wrong." It's modeling a particular situation. The criticism may be that that situation/model isn't very useful, and if SBarr is consistently the better choice, then I'd agree. I'd also expect that SBarr will be nerfed, because if it's better than SBlock in all situations, that's a serious balance error on Blizzard's part.

I'm not sure why you're concluding there's something wrong with ability usage based on the hit/exp values. In a situation where you have enough rage to keep Shield Block up consistently, hit and expertise do have very little value in this model. When you dial hit and expertise back such that your RPS drops beneath the threshold necessary to achieve near-maximum Shield Block uptime, they increase in value dramatically.

In any event, thanks for the feedback. If you have any other comments on the sim, I'd be happy to discuss them. The more we improve it, the better data you guys will have to work with.

[theckhd]

Also, regarding Shield Barrier: It's something I'd like to add. I can already track damage taken in the simulation to calculate an effective Vengeance value, and thus a "size"for Shield Barrier. It's just a matter of coding it, and optimizing the code some if it causes the sim to balloon in time.

The real question is "how do we prioritize it?" For example, do we want to cast it any time we're at X+ rage and have at least Y seconds left on Shield Block? Or do we ignore Shield Block and push it back, risking a drop in uptime (which would further value Hit and Exp)?

I'm not up to date on all of the prot warrior theorycrafting, since I'm fairly busy with paladin stuff. Is there any sort of consensus on how Shield Barrier is supposed to be used? And if not, what are the likely candidates - we could always sim them all out, and see for ourselves what works.

[Tengenstein]

I really think that with optimal uptimes on Shield Block the eternal is a pretty significant increase in TDR over the austere.

Shield Barrier usage come sin many flavours

• Use it exclusively, ignore Shield Block exists
• Use it to prevent capping when Shield Block is on CD
• Use it to cover the periods where Shield block is on CD but the buff isn't up
• Use it decrease spikes in boss damage like psychic drain or impale.

[kopcap]

Theck, you may be missing that Eternal increases both block and crit block for warriors.

Also, idk if this is possible to include some sort of "rage cushion" into your sim, ie while your sim may suggest that the "dramatic jump" in expertise value happens around 6.66 RPS, in reality it will be a fair bit higher because of human error and latency and because we often want a reasonable guarantee, say 95% CI, that we will be able to pool 60 more rage for the next SB before the current one expires. It does not have to be 60, since we have shouts and zerker at our disposal, but we still want to be reasonably sure.

Another thing, if we want to apply Sbar correctly we have to take health into account. Value of Sbar vs SB at 100% health and at 1% health are not the same.

[Quietsch]

Ignoring the meta (by which I assume you mean meta gem?) is basically irrelevant. The armor meta is likely to be much stronger for all tanks anyhow, and a 1% change in block value isn't going to massively shift anything.

Yes, I mean the meta gem - And the SBV/Eternal is much stronger for prot warriors. Of course it isn't that much of a difference, but it still favors mastery (and connected to this also expertise/hit to a certain extent).

Sbarr is ignored for good reason - it's not in the model. The idea was to adapt a simulation I already had (for paladins and SotR) and see what the corollary was for warriors. Namely, if you're trying to maximize Shield Block uptime. That's working under a number of assumptions, first of all that Shield Block is actually your go-to mitigation tool. I've read that due to balancing tweaks, Shield Barrier might actually be better at both 85 and 90. That would require a different model.

That's not what I meant. I'm well aware what you doing there. My 'problem' is simply, that there's no mention (or maybe I'm just missing it) of the fact, that those statvalues are calculated without SBarr. (It's mentioned later in the post, but... dunno, for me it's just inaccurate/-complete like this. And also I've gotten at least 2 mentions, where someone would simply throw around your numbers without seeing, that they're missing SBarr (Again, I know it's intended, but I'm also aware, that they're thus not as accurate as some other persons think they are))

In short, the choice of SBlock over SBarr isn't inherently "wrong." It's modeling a particular situation. The criticism may be that that situation/model isn't very useful, and if SBarr is consistently the better choice, then I'd agree. I'd also expect that SBarr will be nerfed, because if it's better than SBlock in all situations, that's a serious balance error on Blizzard's part.

It isn't better in normal situations (meaning those "I'm a boss and i'm hittin' you!"-situations), but it's still a valuable part of the 'post-SB-Uptime-cap' value of expt/hit (and also a little bit mastery).

I'm not sure why you're concluding there's something wrong with ability usage based on the hit/exp values. In a situation where you have enough rage to keep Shield Block up consistently, hit and expertise do have very little value in this model. When you dial hit and expertise back such that your RPS drops beneath the threshold necessary to achieve near-maximum Shield Block uptime, they increase in value dramatically.

Yes, to be honest I don't know how I got to this conclusion. I've probably looked at wrong numbers or something like this. (Even if I still get higher values for hit/expt (and lower for parry/dodge, relative to mastery) in my sim - But it's basically saying the same thing. Well, it should.)

In any event, thanks for the feedback. If you have any other comments on the sim, I'd be happy to discuss them. The more we improve it, the better data you guys will have to work with.

There's at least one more thing. (At least if I correctly grasp your table) The hit size table.
You've got 34.4% full hits and thus 65.6% mitigated/avoided at 65.9% SB-Uptime. Which is... well, irritating me. Basically, 65.9% of all hits should be at least blocked. There would be 34.1% remaining which could be, in the unlikely event that none of them is blocked or avoided, full hits. (Which would get us to around .6 * 34.1, so ~20% of all hits are full hits) (And if build a quick sim and let it run, it comes roughly to this, with 2.0 swingtimer as well as 1.5)

[Airowird]

I think most people agree that Shield Barrier should (atleast) be used when:
* Your next rage generator would bring you to the cap (maybe +1 rage for the Def Stance rage gain)
* You are below 25-30% HP
And always use it over SB:
* When 2,5 * Stamina is more than 30% * boss's physical DPS
* Or when atleast ~20% of the incoming DPS is magical (weak boss, heavy AoE)
* When you can't block the big attack you're going to get in <2-3s.

I got the exact calculations/formulae for SBar vs SB on my spreadsheet if you're interested in using those. Let me know if you can't find out where all the links go to

Metagem: For Warriors, the Eternal virtually always wins. At 0 Mastery rating you already need ~75K armor to make the Austere worth it. On a normally geared protection Warrior, the minimum armor for Austere quite often goes beyond the armor cap. Again, spreadhseet has all the numbers.

[Airowird]

There's at least one more thing. (At least if I correctly grasp your table) The hit size table.
You've got 34.4% full hits and thus 65.6% mitigated/avoided at 65.9% SB-Uptime. Which is... well, irritating me. Basically, 65.9% of all hits should be at least blocked. There would be 34.1% remaining which could be, in the unlikely event that none of them is blocked or avoided, full hits. (Which would get us to around .6 * 34.1, so ~20% of all hits are full hits) (And if build a quick sim and let it run, it comes roughly to this, with 2.0 swingtimer as well as 1.5)

You're forgetting that avoidance comes before block chance.
With say 20% avoidance, 20% of all incoming attacks are always avoided. Of the remaining 80%, you block 52.72% when SB is up and of the remaining 27.28%, you block say 20%, so 5.456% of all attacks, leaving the rest of the attacks unmitigated.
Total table is then:
Avoidance: 20%
Block: 58.176%
Full hits: 21.824%
Total: 100%

In more abstract form, the chance you block in a random point in a fight is:
Block% = (1-Avoidance) * (SBuptime + (1 - SBuptime) * nonSB_Block%)

[Quietsch]

No, I'm not forgetting it, I'm just talking about guaranteed events - avoid is still only a chance, whereas block while SB is guaranteed, thus you have at least 65.9% of at least blocked damage. (Which is the base of this whole argumentation, as it would still be possible, even if it's highly unlikely, that avoid is 0% - But there have to be 65.9% block)

/e To further illustrate it:
While SB is up, everything is at least blocked (of course after avoidance, but as this is only a chance, and not guaranteed, I ignore it), so those events aren't of any relevance at all, as they can't be full hits.
While SB is not up, those attacks are either avoided, or not avoided and blocked, or not avoided and full hits - which I'm trying to express above.

[Airowird]

Let's see if it's possible...
Working back from the formula above:
Block% = (1-19.3%) * (65.9% + (1 - 65.9%) * B%) = 100% - 19.3% - 34.4%
80.7% * 65.9% + 80.7% * (1 - 65.9%) * B% = 46.3%
80.7% * (1 - 65.9%) * B% = -6.8813%
B% = -25%
So euhm ... yeah, not something that should be happening.

I think the biggest question is if the SB uptime is calculated as 6s * X uses over time, or if it the sim actually shows uptime as Y chance that you have SB up while getting hit.