Brewmaster (Tank) Basic Discussion - Page 9

md096 · 08/19/12, 1:28 AM

I was curious where armor stood in the stat line-up. Primarily curious when it comes to profession bonuses/consumables namely using an haste/armor elixer combo over the agi from flasks. Wanted to find out how mixology would fare using the elixers as I have mats saved up to level this fairly easy profession.

Venyasure · 08/19/12, 2:06 AM

Even if the enchants for dual wielding outweigh a 2h

Do they ?

The duration of the Dancing Steel proc used to be 12 seconds. I ran a few simulations, and they showed 1H was slightly ahead of 2H concerning the average agi gained from the enchants. Yet, it wasn't high enough to outweight the additional base agi you get on a 2H (1Hs have less stats).

Recently, the duration of the proc was apparently reduced to 7 seconds. I modified my simulations, and i ran them with the following parameters :

- 6000 minutes simulations (100 hours each)
- ppm = 1 (like Landslide i think)
- 4k haste
- 2550 hit (hard cap 2H, soft cap 1H)
- 2550 expertise (soft cap)
- 2H with 3.6 weapon speed
- 1H with 2.6 weapon speed
- 80% of the gcd are spent on melee yellow attacks (which are successful) resulting in chances of procs on the main hand

Here is a series of 18 simulations of 100 hours each. The numbers are the average gain of agi due to the enchant :

Now, why do we now have a higher agi for 2H ? I guess it's because, by reducing the duration of the proc (12 to 7 seconds), we reduce the risk of a proc refreshing a previous one, making the firs one less effective since it didn't give a full 12 sec buff. That risk is diminished with 1H : indeed, even if the number of procs is roughly similar, a proc on the off hand doesn't refresh an existing main hand proc, but it stacks with it (thank you Tsuki for this information). Therefore, by reducing the refresh numbers, we diminish the influence of one advantage of 1H. It seems it is diminished enough for 2H to be ahead.

The results are somewhat unstable, especially for 1H (yeah, 595 to 609 for 100 hour simulations is what i call unstable). But they are stable enough for us to conclude that you gain more agi with a 2H.

In addition, the number of Elusive Brew stacks gained with a 2H is 1% higher than with a 1H. You also get more secondary "green" stats on a 2H. Threat generation is nearly identical ; auto-attack seems a little higher for 1H (although i should remake my calculations about it since they date back to when Brewmasters still had Tiger Strikes), but you have more agi (and possibly more crit and haste) on a 2H. I think we can conclude that 2H > 1H, even if it's not way above 1H. Having 1H weapons with better green stats could be better than a 2H with poor secondary stats, especially if it's a fast weapon ; indeed, even if weapon speed does not affect in any way damage reduction, that's only true if we don't consider enchants : a fast weapon will result into less procs on yellow hits.

Finally : Colossus (ppm seems to be 2.4ish). No surprise here, Dancing Steel is better :

(that's only one test but, well, we'll just consider the results would be the same for any test)

And concerning River Song :
agi > dodge
1650 agi for 7 sec > 1650 dodge for 7 sec
we'll suppose ppm is identical
Dancing Steel > River Song

But on some occasions, a dodge proc might be more beneficial than an agi proc. For example, if you consider the Sha of Fear, even if agi provides quite a lot of dodge, you might prefer an actual dodge proc to survive the "Thrash" ability, since, appart from dodge, agi will only give you more Elusive Brew and more healing from Expel Harm and Guard. But that's only a supposition, my guess might be wrong ; maybe gaining more Elusive Brew could be better than waiting fo the dodge proc at the right moment.

Edit :

I was curious where armor stood in the stat line-up

You can find that out using the Excel i linked here. It doesn't display the increased damage reduction gained by increasing armour (like it does for agi, hit,...), but you can modify the "armour" cell and check the results.

By the way, here is the new version adapted to the last build (no more Tiger Strikes, increased EB generation) :

http://www.filedropper.com/brewmastergear2003_1

Don't forget to enable the macros if Excel blocks them !

Taser · 08/19/12, 4:44 AM

It seems that you are right with your math for EB stacks :-)

Let's take your stats:
4000 haste = 9.41% = 20.353% raidbuffed
3000 crit = 5% = 22.945% raidbuffed

3.6 2h = 2.1373 with haste
3 stacks per crit
2.1373 / 0.22945 = 1 crit every 9.315 sec
3 / 9.315 = 0.322 stacks per second

2 x 2.6 1h = 2.16 with haste
1.5 stacks per crit
2.16 / 0.22945 = 1 crit every 9.414 sec
1.5 / 9.414 * 2 = 0.319 stacks per second

But how do you get an equal value for Exp and Hit in your sim if the char is not hitcapped and not exp softcapped? Shouldn't the value of Exp be higher as you are gaining double benefit for dodge and parry?
Real hitchance for whitehits = 100 - 24% glancing - 26.5% miss - 7.5% dodge - 15% parry = 27% left for crit + hit
Real hitchance for yellow hits = 100 - 7.5% miss - 7.5% dodge - 15% parry = 70% left for crit + hit
After being hitcapped + exp-softcapped the value of Exp should still be higher as Exp contributes to both white and yellowhits while hit only affects whitehits.

That leads me to another issue...totally disregarding hit and exp is a bad idea for DW => critcap
100 - 24% glancing - 26.5% miss - 7.5% dodge - 15% parry = 27% left for crit + hit
Just wanted to throw this in as we might see BoK give shuffle even if it misses. In this case there might be someone thinking that hit and exp are no longer of interest apart from KS. It will also influence our EB generation.

Venyasure · 08/19/12, 5:37 AM

Originally Posted by Taser

Shouldn't the value of Exp be higher as you are gaining double benefit for dodge and parry?

You aren't. Expertise doesn't work like in Cataclysm any more. Now, against a boss you have 7.5% miss, 7.5% dodge, and 7.5% parry. Expertise diminishes dodge chance. And it's only when you're above 7.5% expertise (meaning no dodge) that Expertise begins diminishing parry chance. Therefore, the formula for the yellow hit table is something like this :

100% - 7.5% - 7.5% - 7.5% + min(7.5 , hit) + min(15 , exp)

or a more simplified one :

77.5 + hit + exp

The formula is linear with both exp and hit (and with the same coefficient), since exp doesn't decrease both parry and dodge at the same time. I hope my explanation is clear...

100 - 24% glancing - 26.5% miss - 7.5% dodge - 15% parry = 27% left for crit + hit

So, that's only 7.5% parry, we have 34.5% left.

Let's take your stats:
4000 haste = 9.41% = 20.353% raidbuffed
3000 crit = 5% = 22.945% raidbuffed

You can also make a calculation without given stats.

H (capital letter) = haste (%, but let's take it between 0 and 1)
h = haste rating
C = crit chance (between 0 and 1)
W2 = weapon speed (2H)
W1 = weapon speed (1H)
S2 = melee swing (2H)
S1 = you get it
P2 = average number of stacks gained per second with a 2H
P1 = same

S2 = W2 / ((1 + H) * 1.4) (1.4 = Way of the Monk)
S1 = W1 / ((1 + H) * 2) (2 = two weapons)

P2 = C * (W2 / 1.2) / S2 = C * (1.1666 + 0.00002745306952 * h) (haste rating)
P1 = C * (W1 / 1.73) / S1 = C * (1.156069364 + 0.0000272037022 * h)

P2 / P1 = 1.1666 / 1.156069364 = 0.00002745306952 / 0.0000272037022 = about 1.01 (1.0092% more precisely)
So P2 is about 1% higher than P1.

By the way : yes, average EB gained per second does not depend on weapon speed. Weapon speed has only one, single, unique and sole influence over the whole tanking thing : the proc rate of enchants.

Taser · 08/19/12, 6:40 AM

Seems I missed the change with parry. Thanks for the clarification. Last time I did some serious melee theorycrafting was during BC, I have to admit.

Actually I like to calculate with actual numbers. This way it is easier for me to find mistakes.

Taser · 08/20/12, 10:38 AM

Question: Is guard worth casting on CD?

At 60k AP guard absorbs about 132k damage.
To make it a better choice than PB total stagger has to be less than 66k. If you calculate a boss hitting for 200k unmitigated damage = about 120k after mitigation. With 3000 masteryrating and shuffle up 59% of this hit is being staggered = 70800.

In my calcs using guard has either no effect or it even increases our damage taken. Venyasure, did you run your sim with both possibilities: guard on CD / no guard at all?

Venyasure · 08/20/12, 10:58 AM

You can run my Excel program without Guard if you want. In order to do so :

There are three sheets in my Excel file : the first one is for the stats, results, graphs, etc... The third one is just for display purposes on the first sheet. And on the second, you can modify the rate at which you use Keg Smash, Guard, and Expel Harm :

By default, the values are :

-- 1/8 = 0.125 (1 Keg Smash every 8 seconds : you can increase this value to add some "brain lag" on the use of KS on cooldown)
-- 1/30 = 0.0333 (1 Guard every 30 sec : you can modify that too)
-- 1/15 = 0.06666 (1 Expel Harm every 15 sec : you can decrease this one if you think it's not realistic to use EH every 15 seconds without it doing some overhealing)

And yes, the program knows that you have more Chi if you reduce the rate at which you use guard. You can also enter 0 in one cell, which will mean you never use this spell. You can try something like doing 0 KS : hit/exp will then become our poorest stats ^^

You'll notice that by decreasing the value for Guard, the total damage avoided per second diminishes even for a hard hitting boss. So yes, using Guard on cooldown is better than saving Chi for Purifying Brew. But in game, you might want to save Guard for peaks of damage (like the Sha of Fear's "Thrash", or a Madness Impale).

By the way, for the next version of my Excel, i'm planning on adding an option to not take into account the Ox Statue. And about that, the last vesion I uploaded isn't the most up to date : even if it has the increased EB and loss of Tiger Strikes, the damage formula of BM techs is not updated to the new ones. It is on my version on my computer, and rest assured, it changes nearly nothing, since these new damage formulas only have a minor effect on the statue only, which is itself a minor survival increase.

Taser · 08/21/12, 10:18 AM

@venyasure
I tried your suggestion to change the values for guard usage. I was able to reproduce the decrease in damage reduction that you mentioned. After that I had a look at the formulas that you wrote. Just a few remarks/questions:

- it seems you are adding the healing bonus of guard (15%) to the absorb shield of guard
- it would be great to have a counter for PB usage
- it seems that you calculate 8 energy for a missed Jab. I thought that were 10
- do you calculate the cost of 10 energy for a missed KS?
- what is that (0.73*(1/648)) about in your parry calc?
- are calculating with a base dodge of 3.12?
- don't you deduct the 1.5% per level difference = 4.5% from dodge and parry?
- i don't get your vengeance calc. Vengeance=2% of incoming damage before mitigation over the last 20 sec. The formula should look like Boss/vboss*20*0.02

I hope you don't mind that I took a look at your programming.

baneberry.dalaran · 08/21/12, 11:42 AM

With our high value of haste, has anyone simmed Windsong?

Alos, on Beta, windsong can proc multiple times, and with DW I have seen haste, crit and mastery buffs up simultaneously. I have assumed his was a bug, but it has persistent the who beta, even after several reports.

if this remains, experience tells me DW windsong procs> any other enchant, but I have not done the math as I continue to think it will get "fixed"

Also, has anyone tried simming the impact of using glyphed guard to maintain a very high uptime of the +30% self heal combined with expel harm, Zen Sphere, Gift of the Ox? With unglyphed guard the bubble disappears within a few seconds

Venyasure · 08/21/12, 1:19 PM

Hi !

Originally Posted by Taser

@venyasure
- it seems you are adding the healing bonus of guard (15%) to the absorb shield of guard

Indeed, i'm adding the Tiger Palm bonus.

- it would be great to have a counter for PB usage

You mean a cell where you have the time (in seconds) between two PB ?

- it seems that you calculate 8 energy for a missed Jab. I thought that were 10
- do you calculate the cost of 10 energy for a missed KS?

Well, i thought it was 8 for a Jab (20% of the initial cost, i think it's the same for feral), and.. 40 for a KS. Did that change to 10/10 in a recent beta build ? I'm pretty sure that last time i tried, a missed Keg Smash didn't refund any energy and you just lost your 40 energy.

- what is that (0.73*(1/648)) about in your parry calc?

That must be part of the conversion parry rating -> pre dr parry %. The 0.73 might have something to do with the K and C constants, actually i don't really remember how i came up with this formula ^^. I found it here :

Diminishing Returns for Brewmaster Monks | Sacred Duty

It might change when we know the exact values of the constants involved in the calculation.

- are calculating with a base dodge of 3.12?

Yep, the 0.12 comes from the base agi. I used the formula given in the link above.

- don't you deduct the 1.5% per level difference = 4.5% from dodge and parry?

Is it not 0.2 per level in MoP ? Did that change too ? Anyway, i take the 2*0.6% (well, 3*0.6% for boss misses) into account in the final avoidance in my "dmg" function, not in the individual parry and dodge calculation.

- i don't get your vengeance calc. Vengeance=2% of incoming damage before mitigation over the last 20 sec. The formula should look like Boss/vboss*20*0.02

It's not "over the last 20 sec". The buff lasts 20 sec. When you are hit (say by a H hit), your new Vengeance is calculated this way :

NewVeng = 0.02*H + OldVeng*(VengBuffDurationRemaining)/20

For instance, if take a 100k hit every 2 sec, the formula is :

NewVeng = 2000 + OldVeng*((20-2)/20)

A general formula, if S is the melee swing of the boss, would be :

NV = 0.02*H + OV*(20-S)/20

Thus, you've got a series :

a(n) = ((20-S)/20)*a(n-1) + 0.02*H

This series tends to 0.4*H/S = 0.4*BossDPS

You can also solve a stability equation : OldVen = 0.02*H + OldVen*(20-S)/20, you get the same result.

Errr, actually, your "0.02*20*bossdps" works too ^^ since 0.02*20=0.4

Errr², i think i know what you mean : in my program, "boss" is BossDPS, not BossDamgePerHit : "Boss/vBoss" would be "damage per second per second". BossDamagePerHit is Boss*vboss

I hope you don't mind that I took a look at your programming

I'm glad that you did. The more people look at it, the more mistakes we'll find (but i hope there aren't any ^^)

Now, here's quite a big update of my Excel :

http://www.filedropper.com/brewmastergearcopy

It looks like that :

Bux fixes :

- Swift Reflexes couldn't miss ; they can now
- Tiger Palm armor penetration bonus was still 5% per stack ; now it's 10%

New options :

- dual wield / 2H : you can enter 1 for 2H and 0 for 2*1H in the appropriate cell. If you enter 0, you'll need to specify the weapon speed and weapon dps of each of your two weapons. I you choose 2H, you just need to fill in the "main hand" cells. Don't forget to modify the weapon dps in the second line, alongside with your other stats. Note that the 2H/1H option does not take into account the loss of agi (there's less base agi on 2*1H than 2H)

- you can choose which enchant you put on which weapon. I've worked out a formula that is coherent with my simulations of enchants uptime for 2H and 1H, i even tested it for different buff durations. Windsong is not implemented yet.

- options for T14 set bonuses

- the DPS of the Monk is displayed (the second line is not the dps where you would spend all your Chi on Blackout Kick ; you do one BK every 6 sec, and the Chi left is used on PB and Guard)

- sheet 2 :

- you can now choose whether or not the Ox Statue will be taken into account

- once you've done a certain number of Jab, Expel Harm, Keg Smash, Guard and Blackout Kick per second, there's still some space to fill with free Tiger Palms before you're gcd locked. The value 0.8 means 80% of the spare gcd are spent on Tiger Palm

Some remarks :

- 2p set bonus boosts crit quite a lot, it becomes better than haste in some (many ?) cases

- if you don't take the Statue into account, hit/exp falls behind haste/crit in some (many ?) cases

- an amazing remark : yes, 2H Dancing Steel is superior to DW double Dancing Steel. And yes, 2H Dancing Steel is superior to 2H Colossus. But DW double Colossus seems to be better than DW double Dancing Steel (that's because 2H benefits more from Dancing Steel) in some (many ?) cases. And actually... DW double Colossus seems to be better than 2H Dancing Steel in some (many ?) cases. That's true only if you take the Statue into account (DW has a slightly higher dps, not much higher, but sufficiently to boosts the Statue enough for 2H Dancing Steel to be a little behind).

- stat priority is clearly not written in stone : it does depend on gear, on boss dps, and on the options with which you run the program

Also, has anyone tried simming the impact of using glyphed guard to maintain a very high uptime of the +30% self heal

That's an excellent idea for the next version of my program.

has anyone simmed Windsong

As you said, Windsong is a bit strange, possibly bugged. And i've not yet begun working on it. But i will, hopefully before MoP is released.

A few questions :

- does anybody knows the proc chance of Gift of the Ox ? Since i don't know the proc rate, i didn't include it in my program, but i guess it's an important ability, and it cannot be neglected. Taser said "autoattacks with 5.2% and styles with 10%". Is it still true in the current beta build ?

- i wonder whether the ppm of Dancing Steel is really 1, and if that of Colossus is really 2.4.

- *many questions about Windsong*

Astrylian · 08/21/12, 2:15 PM

Avoidance reduction is indeed 1.5% per level in Mists. Bosses negate 4.5% dodge, 4.5% miss, and 4.5% parry. Since miss chance is a flat 3%, you'll never be missed by a boss.

Theck · 08/21/12, 2:34 PM