Mastery, Hit and Expertise Analysis (Prot Warrior)

[kebess]

Hi, guys.

Here's a Mastery, Hit and Expertise analysis.

http://www.youtube.com/watch?v=UvwI2X4J0Mk&feature=plcp

As always, do let me know if you disagree with anything that is said, or would like an in-depth view on any particular section !

[Tengenstein]

Just looked at your notes, on MMO. First of all where are bosses getting a 1.8s swing times from? Secondly your maths for working out the average rage from 6s cycle is wrong as it doesn't take into account any rage from extra revenges from parries/dodges or cycles being cut short due to SnB procs or avatar or shouts, so is rather off. Also calculating the RPS at 630 ilvl is kinda redundant till we get high ilvls, and due to the fact is crit block is on a 2 roll system Mastery's value will inflate as you get more, when running around in your 440ilvls 1% of threat stats will increase your RPS by about 3 times the amount 1% mastery will. Stat values will change as your stats change. Are we even going to see 630 ilvls? tier 14HC stuff is what 514ilvl?, add 39Ilvls on top assuming we get 4 tiers in mop I doubt we'll get up to 570, even with VP bonus ilvls and endboss drops higher ilvl stuff

and of course the whole thing is out of data as of this morning now we can actually have 100% uptime on SB rage permitting. I've been building a little rage calculator for us prot warriors you may be interested in;
15913.1 rage calc.xls

[kebess]

I'm a bit disappointed you didn't understand my maths and ultimately the video, it means my analysis wasn't clear enough.
I could have added a bit more details, but didn't want it to be too long neither.

Anyways, let's clarify things a bit :

Just looked at your notes, on MMO. First of all where are bosses getting a 1.8s swing times from?
Attachment 3460

It's an approximation as the boss swing will vary depending on the specific boss. In the past we have seen most bosses (meaning most of them and not all of them) to have a swing time of about 1.5 sec. When you then applied the 20% attacks speed debuff (TC), you got exactly 1.8 sec swing time. This was the case in DS and FL before that.

In MOP, it will also be the case, with the exception that most bosses will directly have this 1.8 sec swing time as the attack speed debuff have been removed from the game, and thus their attack speed is directly scaled accordingly. This was confirmed by Blizzard some months ago.

Secondly your maths for working out the average rage from 6s cycle is wrong as it doesn't take into account any rage from extra revenges from parries/dodges or cycles being cut short due to SnB procs or avatar or shouts
Attachment 3460

It does take into account Revenge procs from dodges and parries, and it also takes into account S&B procs, this is even why Devastate is there in the first place, as the only way you can get Rage from Devastate is through S&B.

Now, this is how I visioned it for the sake of simplicity :

In my calculations, I have a total of 25% dodge+parry, meaning on average you'll be avoiding only 0.8 attack every 6 sec (or 1 attack every 7.2 sec), IF the boss swing is equal to 1.8 sec.

This means 80% of the time, you'll actually have one extra Revenge to use, and so 80% of 10 Rage = 8 Rage from that.

Do you follow ?

The S&B part is explained in the OP on MMO, feel free to check it out, here.
I will add that, since you're using Devastate at the end of the cycle, it can't be ''cut short'', it will simply increase the Rage generation of the Shield Slam of the next cycle, which is indeed taken into account in my calculations.

Avatar is a 3 minutes CD so no I didn't take this into consideration. I assume it's not something that will be available all the time, or even most of the time, it can only have a 11% uptime.
And Shouts aren't represented neither, as they don't need Hit or Expertise to provide Rage.

Also calculating the RPS at 630 ilvl is kinda redundant till we get high ilvls, and due to the fact is crit block is on a 2 roll system Mastery's value will inflate as you get more, when running around in your 440ilvls 1% of threat stats will increase your RPS by about 3 times the amount 1% mastery will. Stat values will change as your stats change. Are we even going to see 630 ilvls? tier 14HC stuff is what 514ilvl?, add 39Ilvls on top assuming we get 4 tiers in mop I doubt we'll get up to 570, even with VP bonus ilvls and endboss drops higher ilvl stuff
Attachment 3460

I clearly said in the Video that the Mastery hard cap, wasn't reachable, because you CAN NOT have the 630 ilvl gear needed. by the end of the MoP expantion you'll be only at some 530/540 ilvl, If we have 3/4 Tiers.

and of course the whole thing is out of data as of this morning now we can actually have 100% uptime on SB rage permitting. I've been building a little rage calculator for us prot warriors you may be interested in;
Attachment 3460

The current stat of Shield Block in the Beta is bugged. Right now you can have 2 charges of Shield Block in 9 seconds, which is absolutely not intended, because this gives technically a ''120%'' uptime, which is a bit silly.

I hope it's clearer now. In any case, let me know if you disagree with something else !

[kebess]

PS : Thanks for that Spreadsheet I'll check it out.

[Fetzie]

I clearly said in the Video that the Mastery hard cap, wasn't reachable, because you CAN NOT have the 630 ilvl gear needed. by the end of the MoP expantion you'll be only at some 530/540 ilvl, If we have 3/4 Tiers.

Heroic T14 is item level 509. Heroic T17 will be something around 600.
http://mop.wowhead.com/item=87104

[Tengenstein]

so let me get this right.... you're video and its recomendation are based off something we're never going to be able to achieve? urgh

I get how you worked out your cylical rage, its just simplistic to the point of being flawed and given how easy it was for me to generate the markov chains and their probabilities i really don't see why you wouldn't choose a lass flawed model

Also, I was under the impression most bosses currently had a 2 second swing timer which was increased to 2.4 with the no longer present dehaste debuff.

ANd whilse SB may be bugged, I'll calculate for the present state, Deep wounds munching is a bugg, but it something to take into account for as long as it exists. Of course the letting you have 120% uptime on SB actuall makes it use a little more strategic, since you have to time your SB, rather like how most clases try to avoid refeshing dots early where possible. a very good tank would require 10RPS and keep SB up 100% of the time if he timed it prefectly a less skilled tank would clip SBs and thus end up needing more RPS to achieve the same effect, taking away from RPS that could be used to put up Sbar for additional survivability or HS/CLeave for additional damage.

[kebess]

Heroic T14 is item level 509. Heroic T17 will be something around 600.
http://mop.wowhead.com/item=87104

Are sure this is the correct Ilvl of the T14 ? It appears to be higher than every thing we've seen so far.

here is how I came to the 530-540 conclusion :

In Wrath, the ilvl of a Heroic dungeon item was 187, and the ilvl of a the 1st Tier normal item was 200, and 213 for the 25 (or Heroic) version.
Difference between HC dungeon and 1st HC raid Tier = 26 ilvls.

In Cataclysm, the ilvl of the first heroic dungeon was 346 ilvl, and the ilvl of the 1st Tier normal was 359, and 372 for the Heroic version.
Difference between first HC dungeon and 1st HC raid Tier = 26 ilvls.

Following this pattern, the ilvl for the HC Tier 14 items should be 489 ilvl , since the ilvl of the first heroic dungeon in MoP, are 463 ilvl == > 463 + 26 = 489.

So I'm a bit surprised by this.

[Tengenstein]

He's correct. Remember there isn;t a difference between a level90 dungeon and a level 90 HC dungeon. They're the same thing.

[Fetzie]

Ah yes, but this time we have raid finder messing things up.

According to Wowhead:
Raid finder is 483: http://mop.wowhead.com/item=86688
normal is 496: http://mop.wowhead.com/item=85348
heroic is 509: http://mop.wowhead.com/item=87104

actually, I think the raids drop different item levels.

see this item from elegon (mogoshan vaults)

LFR: http://mop.wowhead.com/item=86795
normal: http://mop.wowhead.com/item=86136
heroic: http://mop.wowhead.com/item=87058

[kebess]

so let me get this right.... you're video and its recomendation are based off something we're never going to be able to achieve? urgh

Again, no. Was I that vague ? : (

This hard cap was merely there to show that there was, currently (as it may change in a future build), no limit to the amount of Mastery we could stack, since we can't actually reach the theoretical limit.

So nothing is based on the Hard cap. It's a simply illustration.

I get how you worked out your cylical rage, its just simplistic to the point of being flawed

How is it flawed ? I might redo a more complete one tomorrow, but I still don't see how it's flawed, as again, its only purpose was to illustrate that 22.5% of the total RPS of SS, Rev, and Dev, wasn't close to be high enough to make Hit and Expertise more valuable than Mastery.

Also, I was under the impression most bosses currently had a 2 second swing timer which was increased to 2.4 with the no longer present dehaste debuff.

Most bosses I've fought so far in beta have had 1.8 sec swing time. I'll try to add logs of this tomorrow.

[kebess]

Ah yes, but this time we have raid finder messing things up.

According to Wowhead:
Raid finder is 483: http://mop.wowhead.com/item=86688
normal is 496: http://mop.wowhead.com/item=85348
heroic is 509: http://mop.wowhead.com/item=87104

actually, I think the raids drop different item levels.

see this item from elegon (mogoshan vaults)

LFR: http://mop.wowhead.com/item=86795
normal: http://mop.wowhead.com/item=86136
heroic: http://mop.wowhead.com/item=87058

Indeed, thanks for seeing this. I'll correct it with an annotation.

This means we might get more closer to the Hard cap than what I initial thought, but we still can't reach it though. Even with this ilvl increase we will still be at around 589 ilvl, after Tier 16.
If there happens to be a Tier 17, we might get really close, with a 619 ilvl gear, yes.

[Tengenstein]

yeah you where that vague the video gave me impression that you think mastery is a superior RPS stat, though you want may want to get threat stats to smoove out your rotation. that just is not true until you reach higher levels of mastery or threat stat caps.

Its flawed beacuse you fail to understand we don't always have 6 second cycles and thus your method for calculating Rage per GCD is incorrect. You say a Dev is worth 0.25 Rage but it isn't. Let me break this down into Markov chains for you, assuming we don't have Rev to deal with (becuase that really complicates things), or rotation is mad eup of discrete repetitive units (cycles)

A.dev>dev>dev>SS
B.dev>dev>dev>SnB
C.dev>dev>snb
D.dev>snb

Probability of A occuring is 0.343 (0.7^3), B is 0.147(0.3*0.7^2 ), C is 0.21 (0.3*0.7), and D is 0.3. A is worth 15 rage over 6 seconds, B is 20 rage over six seconds, A is 2.5RPS, B is 3.33RPS, C is 4.44RPS and D is 6.66RPS, this means that on average our RPS is 4.28RPS, and our average cycles is actually 4.785, and contains 2.19 Devs, so assuming each cycle contains one 15 rage shield slam whatever is left over is the avarage RPS of devastate. 15/4.785 gives our SS a RPS ~3.135, 4.785-3.135 is 1.6RPS from devastate, and on average we ahve 2.19 devs per cycle, so after that division Dev is worth 0.731RPS. and it gets worse as you as you add in more revenges, as it gets harder and harder to shorten the cycle time which is the biggest factor in in increasing the RPS of Dev.

Additionally your rage from revenge is wrong, "1 Rev (66% of the time -- Rev as a 9 sec CD) => 66% of 10 Rage = 6.6 Rage" is fine, except you don't account for the other 33% of the time, i assume you don't intend for us to spend the GCD doing nothing? same goes for your other Revenge on proc maths.

Essentially what you need to do to calculate avearage rage Rage from rotation is calculate the rage from every possible cycle and then caclulate the probabilty of each cycle occuring and then go back and do it again as missed Devs don't proc SnB (after you have a little cry)

[kebess]

Essentially what you need to do to calculate avearage rage Rage from rotation is calculate the rage from every possible cycle and then caclulate the probabilty of each cycle occuring and then go back and do it again as missed Devs don't proc SnB (after you have a little cry)

You do agree it is far from being as simple as you said earlier then, ; )
Could you finish your calculations and give an approximative RPS value for an average cycle of those 3 attacks ?

So I can compare it to my 5.18 RPS, and see how far I am, and make sure it doesn't change my final argument ?

[Tengenstein]

I didn't say it was simple I said it was easy. Its all just multiplication, and you could just use said caclulator thats been up for a month here, plug the numbers in and get an answer.

Assuming you have 25% avoidance, 23% mastery and 0%hit&exp and a 1.8s swing time i get 5.151RPS
Assuming you have 25% avoidance, 23% mastery and capped for hit&exp and a 1.8s swing time i get 6.983RPS
Assuming you have 25% avoidance, 45.5% mastery and 0%hit&exp and a 1.8s swing time i get 6.435RPS

also having had my little cry15913.2 rage calc.xls

[kebess]

Ok but, without the Mastery, I only need the RPS from SS, Rev, and Dev, sperate from the RPS from Mastery.

So I can compare it to the RPS generated from Mastery through cirtical blocks, calculated seprately.

So in the end, all you do is : [0.225 * (RPS of SS, Rev and Dev)] - [RPS of 28% cirtical block] = ?

I got 0.4 RPS with my rough claculations, so it was low enough to make my point.

[kebess]

I'm checking out that Rage sheet of yours right now.

[Tengenstein]

So in the end, all you do is : [0.225 * (RPS of SS, Rev and Dev)] - [RPS of 28% cirtical block] = ?

No. Because less of your attacks connect you get less rage, proportional to the amount of hit/expertise you're lacking, but you also lose rage because you get less SnB procs, so you lose even more rage for not having the +5 rage procs, but you also have less shield slams in total, which is even more Rage loss, and because of the rage loss you start having a negative effect on Shield block uptimes, and because you have less shield block uptime you have less critical blocks, and because you have less critical blocks you have less enrage procs, which is yet another loss of RPS.

I've given you a calculator that takes into account all of those factors, You don't have to reinvent the wheel; you can see all of my workings in it.

EDIT updated the calculator again.

[kebess]

Ok, I get a difference of 0.548 RPS (up from my previous 0.4), when I take into account the chain reaction effect you just described. Which results in reducing Shield Block uptime a bit more with 0% Hit&Exp.

Now, the question becomes, is 0.548 RPS still low enough to make Mastery still more valuable, through the extra survivability we get from blocks and critical blocks, when we actually decide to stack it, and neglect Hit and Expertise.

I'll be crunching some numbers later on today to confirm this.

[Tengenstein]

You do realise that becuase Hit, expertise, Dodge, parry, and mastery all give rage gen the relative value of any one of them will fluctuate in accordance with the relative amounts of all the others? and that this also holds true for the direct damage reductions ofered buy such stats. Further more since stam, gives AP and strength gives both parry and AP, and as Sbar damage absorbtion is based off AP you start running into what i would consider "horrendous maths".

Good luck.

[kopcap]

Teng, back to the idea that I started explaining before.

Lets say we have a prot warrior Frodo with an X mount of rating that can be freely redistributed any way we like between stam, dodge, parry, mastery, expertise and hit (purely for illustration sake, we can apply any reasonable constraints later on). Frodo plays optimally.

Lets say we define CI as a confidence interval of having 60 or more rage at the end of a 9 sec cycle.

Using existing model we can derive distribution N of stats (stam, parry, dodge, mastery, hit and exp) that:
a) fits CI
b) minimizes the amount of rating used to achieve this, lets call it Y

This distribution N represents the minimum amount of rating that statisfies SB uptime within predefined CI and other arbitrary constraints of our choice, ie 300k HP.

After this we are left with (X-Y) amount of rating that can be used in one (or a combination) of three primary ways:
1) increasing EHP through stam stacking
2) increasing Damage Reduction through avoid/mastery stacking
3) increasing Damage Reduction and EHP (sorta) though SBar uptime and threat stats

Now, what interests me is given distribution N to be able to translate each point of rating spent on a threat stat after Y threshold into Sbar uptime. So we can compare its DR benefits vs the avoidance/mastery path and visualize its stamina cost.