Updated for 5.4
Updated for 5.4
Nice calc, but why aren't you taking into account Charges and Shouts ?
I'm not taking into account charge becuase most tanks do not charge on CD. SHouts aren't modeled becuase for the most part the RPS of a shout a will depend on which abilitly you use replace or delay using when casting a shout. which is kind athe same reason i don't take into account the RPS loss of having to keep WB up via TC. casting shout on CD is not just +0.3RPS, its 0.3-RPS of the abilty you would otherwise use,and if you delay the shout itself, which you probably should if you have Snb, or Rev up, especially so if avatar is up, you start diminishing the RPS of the shout aswell.
Edit: updated the sheet to include DR from armour.
Just quickly looking over the spreadsheet and a few questions.
Is the base 5% miss we get gone now?
Also cell b48 looks like you are missing () if I understand what you are trying to do.
Trying to figure out exactly what is going on in here but it seems like their is a loop somewhere because if you enter the same stat value multiple times the RPS / DR / SB % values change slightly.
Thanks for spotting those.
I have missed the 4.4% chance bosses have to miss. and there is a loop because average SB uptime affects your Critical block rate which affects your RPS which affects your average SB uptime.
Miss was changed with MoP, you've got 3% baseline and 1.5% reduction per enemy level over player level, so you'd have -1.5% versus bosslevel. (And if i'm reading the only bluepost i seem to be able to find at the moment correctly, the generell reduction for avoidance is now 1.5% per level)
Ouch, bosses have a 12% higher chance to hit you? That's brutal...
Well, it would be 3*1,5*3 = 13,5 (or even 18, if you'd want to add block...) :>
But i don't know if the reduction is really like this. It would be quite extrem imho.
If base miss chance is 3% the boss couldn't really ignore 4.5% of your miss chance. Would make it 4.5 from dodge 4.5 from parry and 3 from hit. I wonder if this does affect block... that would be horrible.
@Quietsch; Thank you for you PM, i can only respond to a PM every 2 hours, so please forgive me for responding here. If you unhide the top of my sheet you'll see it doesn't assume all chains are six seconds, and it also doesn't exclude rage from avoidance procced revenge in calculating the base rage, Hopefully it should look familair. I'm not quite sure where my calc is differing from yours, but if you set the expertise/hit to capped (10k rating in each) and the set chain caster to yes, cell I17 should give you the same number as your calculator. It doesn't. we are getting the same average cycle length (J17), and same probabilites for a SS/Dev only rotation, i'm fairly certain you're right, but the PM kinda garbled the formatting so theres alot of "<>" in places where i think you meant to start a new line if you would be so kind try to nail down where I've gone wrong in the spread sheet.
thanks for head up about the miss/avoidance redcution, Knew i should have been lazier about responding to takethecakes miss chance thing.
@Takethecake i can't imagine why it wouldn't affect block. block suppression happens now, why would it be a special case in MoP
According to the bluepost, it affects block. (But well, who knows, if it's really like this.)(And you got a point there, at 3% miss - I'm still living in pre-MoP-avoidance-times :P)
Block isn't, at least as warrior, this much of a problem, as you got SB (... well, if SB were flat block, not counting your baseblock so including reduction you weren't capped anymore, that would be horrible.).
(those <> where meant to be... erm... equivalence symbols (I don't know, if there's a specific word for those))
/e just to clarifly it: I have only looked a (very) short time at those chains with options other than hit & expt cap + chain caster Yes, so i have nearly no idea, if the results they give are correct or not, my whole post is only directed at those settings. (My guess is, that they are a bit off, because of those already mentioned problems, but as all are 6s long, it is probably a relative lower relative error)(I'll look at it... either tomorrow, or later)
And now i've finally understood the problem of your calculation.
(Everything assuming hit and expertise cap, also ignoring latency and so on, as well as on a chain-casting boss (and revenge ignored, which is another problem in your sheet, as it still uses a GCD (-> you can use less Dev, so less (possible) SnB-proccs)))
First, the values for the chains (same as in the sheet):
(1) SS - Dev - SnB-SS: 0.3, 20 Rage, 3 Seconds
(2) SS - Dev - Dev - SnB-SS: 0.7*0.3 = 0.21, 20 Rage, 4.5 Seconds
(3) SS - Dev - Dev - Dev - SnB-SS: 0.7*0.7*0.3 = 0.147, 20 Rage, 6 Seconds
(4) SS - Dev - Dev - Dev - SS: 0.7*0.7*0.7 = 0.343, 15 Rage, 6 Seconds
Your sheet does the following:
RPS = 0.3*20/3 + 0.21*20/4.5 + 0.147*20/6 + 0.343*15/6 = 20*(0.3*1/3 + 0.21*2/9 + 0.147*1/6 + 0.343*1/8) (the second term is so to say Rage * average chainduration)
Basically, every style has a certain RPS, so total RPS is the sum of all style-RPS multiplied with the appropriate percentage.
The problem is, that it doesn't account for the... fitting duration of each chain. The chain duration in this formula would be 1/(0.3*(1/3) + 0.21*(1/4.5) + 0.147*(1/6) + 0.343*(1/8)), which is not equal to the correct duration (And also does it grant to much rage per chain, if you'd ... shift the other RPS formula into the same 'format', it would bei 0,91425*20 Rage per chain)(To be honest I'm still not quite... well, sure, as well as able to write it understandable, why the formula shows the results it does, the only thing i'm able to say is, that it's wrong :P (It would probably require same GCD-lenghts, to be working or something like that - which is quite a bit of required work)).
RPS = (20*0,3 + 20*0,21 + 20*0,147 + 15*0,343)/(0,3*3+0,21*4,5+0,147*6+0,343*6) ( = average Rage per chain divided by average chain duration)
One thing that really intrigues me on that GC post is this part :
This means we simply aren't gonna be missed anymore by bosses, and we are coming from a 0.2% per lvl avoidance stat bypass to a 1.5% per lvl. This is a 750% increase. How is this justified ?Quote:
Secondly, players have been curious about the combat table chances against players of boss mobs. What is described in that blog post is indeed accurate for a boss attacking a player. That does mean that a player has a base 3% chance to be missed, 3% chance to dodge (if they know Dodge), 3% chance to parry (if they know Parry), and 3% chance to block (if they know Block). That chance is indeed reduced by 1.5% per level difference, so a 92+ will never miss a 90. So to use a simple example, a monk with no parry on gear or other buffs has a 3% base parry chance, plus 5% from Swift Reflexes, for a total character sheet parry chance of 8%. A level 93 boss attacking that Monk will get parried 3.5% of the time. Hope that clears things up.
I for example use Charge coupled with my HL (glyphed to a 30sec CD). I place that combo between two GCDs, which prevents altering my GCD rotation.
And so more or less every 30 sec, I have 1 Charge that grants an additional 30 Rage, equal to very roughly, +1 RPS to my overall avg RPS amount, only through Charge.
Shouts are on the GCD so can't proceed as such. Since Shouts provide 20 Rage, using them and delaying a non SnB SS, will still provide more Rage, and using it instead of a Revenge will provide even some more Rage.
In fact, unless I'm missing something, you'll never lose Rage per GCD by using a Shout instead of an SS, Rev or a Dev (unless Avatar is active, which will be only 20/1.8 % (11.1%) of the time).
Again, I understand that this complicates things for you, but by not taking those into account, you can't make any solid RPS conclusions based on that calc.
The problem with charge is, that it's not simply a cooldown, but has even moar requirements, so it's relatively hard to model it (at least approximately) correctly in an not-totally-complicated calculator (Of course, if you assume, you'd never postpone your GCD, it's totally simple - but... i'd highly doubt that.). (But with this simplified model, it would be exactly 1 RPS (Shouldn't charge be 20 Rage, so 2/3 RPS?) (or rather, it would be exactly 30(20?)/x, whereas x your Chargefrequency is))
You lose Rage if you use a shout instead of an SnB-SS. (Simply, because the relative amount by which the SS CD is postponed is greater, than for a shout)
The optimal use for shouts is instead of the 3rd Dev in a row. (Of course every other place with the exception of SnB-SS is also increased RPS, but (much) less)
/e @ Baseavoidance change: It depends on mechanics and so on, so you can't simply say 'It isn't justified' (I don't want to imply you are like this). Overall it can make sense or not. As the mechanics of all tank-classes are changed, it's somewhat reasonable, to also change those 'basic' mechanics. (It's the same with block as 2nd roll. It isn't justified or not justified, rather it's 'useful/target-aimed', or not.)
Combining Charge to Intervene to have more uses, on the other hand, is a bit more trickier to place in between 2 GCDs without any delay.
I thought roughly 1 RPS, because you might delay the Charge a sec or 2 while timing it so it is placed exactly in between 2 GCDs, and thus bringing that 30 sec CD to something like 31 or 32 sec. But for simplicity sake, it can be rounded up to 30 sec, yes.
You might indeed want to place your Shout right after that SnB proc as you said, which would inevitably increase your RPS.
I see - but it's missing on wowhead (or i'm simply getting old/blind... :/).Quote:
I also use the glyphed Charge, which provides 30 Rage instead of 20.
In his kind of simulator it can't happen. Also, even in reality it should nearly never happen (or at least not, if your latency is good enough to not miss an SnB proc for an entire GCD).Quote:
I can see how that can happen, but the chance of it happening is also much slimmer than the chance of it not happening.
You might indeed want to place your Shout right after that SnB proc as you said, which would inevitably increase your RPS.
Also, you don't want to use it right after SnB, but right before an _normal_ SS. (Right after SnB you'd use either revenge or devastate - revenge with 10 rage, and if it's postponed something between 10/9*1,5 Rage up to... well, in the worst case 10 Rage loss, devastate with... dunno, 2-4 Rage (probably the higher one) i'd guess... loss. (I'm too lazy to calculate it right now, sorry))
I think i get it.. lemme up load another spreadsheet and you can check if i actually got what you meant. And yes this would be so much easier if we actually had the vocabulary to understand what we' meant.
on the whole Rev against chain caster. yes it is poorly modeled, I don't know how to do it better, however being that mastery/parry/dodge all are completely worthless against chain casters i'm not to fussed about it, the stat priorty is just hit=exp>stam and not particularily interesting, I suppose we could just append +10/9-rps/9 for chain casting.
since Revenge only gives 10 Rage, and Devastate in the best case scenario would proc a second SnB, that could be used right after.
But since SnB+Shout => 2 GCD, and SnB+Dev+SnB => 3 GCD, you'd still be gaining more RPS with the Shout combo.
Of course, I see how you could get even more Rage by placing your Shout on the 4th GCD of the rotation, instead of the 2nd Dev, if there isn't a second SnB proc.
But in any case, using the Shout after the 1st SnB (whether it's right after or on the 4th GCD) should always provide more Rage than not using it at all. Do you agree, or I'm missing something ?
Erm... You're missing quite a few points (to answer your question(s): Yes, it grants even more Rage. Also, using Shout grants equal or more Rage than not using it (and in most cases more)). Especially that every rage-granting style got a CD. And also that we've got a potentially infinite chain of styles. (Of course, in a fight, it would end at some point, but in our 'optimal', theoretical model it doesn't).
In your SnB - Shout - Dev - SnB case, you'd get 60 Rage over 6s. In the case of SnB - Dev - SnB - Shout you'd get 60 Rage over 6s.
But: In your case, at the end of the chain your SS has 6s CD. In 'my' case i've got 4.5s CD. So basically, if there wasn't another SnB-Proc, it would be SnB - Shout - Dev - SnB - Dev - Dev - Dev - SS for 75 Rage over 12s, versus SnB - Dev - SnB - Dev - Dev - Shout - SS for 75 Rage over 10.5s. /e fixed stupid mistake...
It's all based on the fact, that a CD delay for a Shout has a much lower influence of the Rage generation from the shout (to be exact (if i'm not missing something besides the Glyph which is imho, as prot, (nearly) worthless, as you want to use as many GCDs as possible to generate Rage, which is opposed to a glyph which grants you the same amount of rage but over 2 GCDs, so the cooldown of Shouts is 60s) it's 20/60*1,5 Rage lost per GCD, whereas SS is 15/6*1,5 Rage lost per GCD (or even more with SnB))
/e Just an addition: There are times, when your method would grant better Results.
For example, if you got the case that _every_ Dev causes SnB.
In 60 GCDs there would be, with my method:
30 Dev, 30 SnB
your method: 30 Dev, 29 SnB, 1 Shout (Dev at GCD 0, so... durr. It's obvious the same. The more... well meaningful example is starting with SnB (or SS): I'd have 30 SnB, 30 Dev, while you'd have 30 SnB, 1 Shout, 29 Dev, so as long as we don't look at the next GCD, you'd have 'won')
(Well, it's a example, with a chance of ... 0,3^30 = 0,0000000000000002058911, so a small probability. :P)
What I mean is, in both scenario, you will gain some more Rage, than if you didn't use your Shout at all.
I thought in you previous post you were saying you could lose Rage by, let's say doing : SnB - Shout - Rev - Dev, instead of , SnB - Rev - Dev - SnB.
In the second scenario, your SS is on CD after the rotation, which is worse than in the Shout rotation, since it's up right after.
But I totally agree that an even more optimal way of using your Shout would be on the 4th GCD.