# [H2P][7.1.5] Raid Stat Weights and Discussion

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Timmylol** » 23 Feb 2017, 21:03

Jockel wrote:Mafuyu wrote:More haste gives you more room for errors.

Why is that`? I feel like having less haste would give you less dependence on maximizing Void Forms, thus giving more 4pcs buffs. Also a lower castspeed kinda should forgive more milliseconds of doing kind of nothing, since with more haste they would count more? Am I so far off in my head about this?

Because lingering insanity carries over to a new void form. Thus, the longer you stay in voidform, the higher LI bonus you have, enabling you to enter the next voidform faster and with bonus haste, creating a snowball effect.

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Jockel** » 27 Feb 2017, 10:07

wSorry, but this doesn't really answer the question for me :( The downtime of VF might be higher for less haste, but that may be compensated by higher damage overall.

Lets just assume 2 scenarios:

A: point blank 12k haste

B: less haste probably 9-10k

Further assume:

-Both scenarios do the same amount of damage, compensated by mastery in secenario B.

-Both scenarios scale quite similar in single target and multi target fights.

Why is scenario A better than scenario B? Why does A forgive more errors?

Lets just assume 2 scenarios:

A: point blank 12k haste

B: less haste probably 9-10k

Further assume:

-Both scenarios do the same amount of damage, compensated by mastery in secenario B.

-Both scenarios scale quite similar in single target and multi target fights.

Why is scenario A better than scenario B? Why does A forgive more errors?

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Mafuyu** » 27 Feb 2017, 11:19

More haste = more casts in the same amount of time.

If you miss a spell with low haste you are punished hard.

Let's say you miss a MB because for whatever reason you casted a SW:P instead.

So, you missed 11 insanity. Which results in a 1 gcd delayed VF.

Now how much is this? Somewhere between 1,5 and 0,75s.

1,5s and 0,75s are not realistic since you are neither at 0% haste nor at 100%. Just for the sake of the example.

This adds up during a fight.

Now let's take your example.

12k haste and 9k haste.

Let's assume you are NOT gaining ANY haste during a 300s fight.

375 haste = 1%

=> 9k haste = 24%

=> 12k haste = 32%

Now the gcd:

9k haste: 1,5s/1,24 ~ 1,21s

12k haste: 1,5s/1,32 ~ 1,14s

Now the number of casts:

9k: 300s/1,21s = 247,93s => 247 casts

12k: 300s/1,14s = 263,16s => 263 casts.

So you get 16 more casts. Therefore, you have a 16-cast error space compared to the 9k haste.

This is not about dps or dmg it is just about having more room for errors as already told by me and others.

Now when it comes to dmg you are underestimating the dmg from mass hysteria.

Higher VF stacks will result in more dmg for our dots.

You are just looking at VB dmg but this is the wrong approach.

Remember how hard they ticked when we were at high stacks in s2m.

Even now look at botanist fights where spriest with s2m have a 3 million dps spike.

If you miss a spell with low haste you are punished hard.

Let's say you miss a MB because for whatever reason you casted a SW:P instead.

So, you missed 11 insanity. Which results in a 1 gcd delayed VF.

Now how much is this? Somewhere between 1,5 and 0,75s.

1,5s and 0,75s are not realistic since you are neither at 0% haste nor at 100%. Just for the sake of the example.

This adds up during a fight.

Now let's take your example.

12k haste and 9k haste.

Let's assume you are NOT gaining ANY haste during a 300s fight.

375 haste = 1%

=> 9k haste = 24%

=> 12k haste = 32%

Now the gcd:

9k haste: 1,5s/1,24 ~ 1,21s

12k haste: 1,5s/1,32 ~ 1,14s

Now the number of casts:

9k: 300s/1,21s = 247,93s => 247 casts

12k: 300s/1,14s = 263,16s => 263 casts.

So you get 16 more casts. Therefore, you have a 16-cast error space compared to the 9k haste.

This is not about dps or dmg it is just about having more room for errors as already told by me and others.

Now when it comes to dmg you are underestimating the dmg from mass hysteria.

Higher VF stacks will result in more dmg for our dots.

You are just looking at VB dmg but this is the wrong approach.

Remember how hard they ticked when we were at high stacks in s2m.

Even now look at botanist fights where spriest with s2m have a 3 million dps spike.

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Later** » 07 Mar 2017, 20:38

I know people say not to sim yourself as H2P theorycrafting methodology is more representative of theoretical dps.

I still sim anyway because we should always double-check things if possible and I do have a question regarding haste vs mastery.

I have 11k haste and 11k mastery (prydaz + unstable arcano), so I'm not sure whether I should be gemming/enchanting haste or mastery. Currently its all mastery gems and enchants, so switching to haste would drop my mastery to 10k+ but haste to nearly 12k. Sims say go haste, H2P says go mastery...

Furthermore, earlier in 7.1.5, the top spriests were going mastery at the expense of haste, but now I see most of them switching to haste instead. Granted, most of them do not have the level of haste that I do, so that may be a factor.

For me, I try to get as much information from various sources. Mostly that entails 3 things:

1. Top parses from warcraftlogs

2. Sims

3. Theorycrafting community (H2P)

Generally, these are pretty consistent in their conclusions, and should 1 and 3 conflict with my sims, I'll go with 1 & 3, since top players and the TC community are way better at math than I am and are probably doing the sims better.

In this case though, there's quite a lot of conflicting information so I'm not too sure.

My sims (Patchwerk-style) still significantly value Haste although I have 11k haste AND Erratic Metronome (I am SL and have the 4pc). Do I then follow the recommended stat weights in the H2P post (which values mastery more) or do I use my sims (which values haste more)?

While I'm certain I'll use the stat weights here for most fights in NH, I'm wondering if I should keep a second, haste heavy set for single-target fights like Krosus and Trillax. (although for mythic Trillax, do I still go haste or prioritise mastery like here?)

I still sim anyway because we should always double-check things if possible and I do have a question regarding haste vs mastery.

I have 11k haste and 11k mastery (prydaz + unstable arcano), so I'm not sure whether I should be gemming/enchanting haste or mastery. Currently its all mastery gems and enchants, so switching to haste would drop my mastery to 10k+ but haste to nearly 12k. Sims say go haste, H2P says go mastery...

Furthermore, earlier in 7.1.5, the top spriests were going mastery at the expense of haste, but now I see most of them switching to haste instead. Granted, most of them do not have the level of haste that I do, so that may be a factor.

For me, I try to get as much information from various sources. Mostly that entails 3 things:

1. Top parses from warcraftlogs

2. Sims

3. Theorycrafting community (H2P)

Generally, these are pretty consistent in their conclusions, and should 1 and 3 conflict with my sims, I'll go with 1 & 3, since top players and the TC community are way better at math than I am and are probably doing the sims better.

In this case though, there's quite a lot of conflicting information so I'm not too sure.

My sims (Patchwerk-style) still significantly value Haste although I have 11k haste AND Erratic Metronome (I am SL and have the 4pc). Do I then follow the recommended stat weights in the H2P post (which values mastery more) or do I use my sims (which values haste more)?

While I'm certain I'll use the stat weights here for most fights in NH, I'm wondering if I should keep a second, haste heavy set for single-target fights like Krosus and Trillax. (although for mythic Trillax, do I still go haste or prioritise mastery like here?)

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Elvenbane** » 08 Mar 2017, 09:39

Patchwerk sims (by themselves) are useless for stat weights as they do not represent current in-game scenarios. If you want something more accurate for Krosus sim light movement instead.

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Anshlun** » 08 Mar 2017, 14:08

Later wrote:My sims (Patchwerk-style) still significantly value Haste although I have 11k haste AND Erratic Metronome (I am SL and have the 4pc). Do I then follow the recommended stat weights in the H2P post (which values mastery more) or do I use my sims (which values haste more)?

While I'm certain I'll use the stat weights here for most fights in NH, I'm wondering if I should keep a second, haste heavy set for single-target fights like Krosus and Trillax. (although for mythic Trillax, do I still go haste or prioritise mastery like here?)

Mastery value is increased when you sim more enemies because you start shifting % of your damage that was being done by Mind Blast/Mind Flay to DoTs, which is the reason why mastery is so good in Nighthold, since it has so many adds and cleave possibilities. That and the 4pc bonus.

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Ohydra** » 22 Mar 2017, 11:30

Hi Anshlun - thanks for the post!

Correct me on my thinking if I am wrong - If I want to have the most accurate stat weights for NH raiding on my specific shadow priest running the standard LotV/PI/SL/LI/TOF build I should:

1) Use the LotV_PI_SL_LI_TOF (haste vs. stat weight) graph for reference

2) Find the x value of haste that my character is currently at

3) Determine the most accurate stat weights for my gear (obviously this may change every upgrade i get)

Could you please provide a link to download the Haste vs stat weight LotV_PI_SL_LI_TOF graph?

Correct me on my thinking if I am wrong - If I want to have the most accurate stat weights for NH raiding on my specific shadow priest running the standard LotV/PI/SL/LI/TOF build I should:

1) Use the LotV_PI_SL_LI_TOF (haste vs. stat weight) graph for reference

2) Find the x value of haste that my character is currently at

3) Determine the most accurate stat weights for my gear (obviously this may change every upgrade i get)

Could you please provide a link to download the Haste vs stat weight LotV_PI_SL_LI_TOF graph?

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **faare** » 03 May 2017, 04:49

@Ohydra

I asked for the graph aswell with no success on the discord so I tried my best to "replicate it".

I added points at every 500 increment in haste, and then used the tools within Excel to model a trendline following them as close as possible (6th degree polynomial was the best fit).

Excel remodelling : http://puu.sh/vEazG/f17bfe9b4a.png

Mastery : y = 2E-23x6 - 2E-18x5 + 4E-14x4 - 6E-10x3 + 5E-06x2 - 0,0222x + 40,08 with R² = 0,986

Haste : y = -3E-22x6 + 2E-17x5 - 5E-13x4 + 7E-09x3 - 6E-05x2 + 0,2218x - 366,83

R² = 0,9968

Alternatively instead of taking these to compute the best stat mix given a stat budget, you can do it more simply with a table.

Here is the full xls : http://www28.zippyshare.com/v/YfHegz49/file.html

If anyone can direct me to an uploader that looks less fishy than that, I can totally reupload.

With this ideally you want to use Excel solver (downloadable but not included in it by default as far as I know).

http://puu.sh/vEaoM/6147ff2385.png

Objective is to maximize score by changing variable cells "Haste" and "Mastery".

This looks great and all but in the end all I just said is useless, because of the "loop" phenomena that takes place.

The "score" you want to optimize is a function of stat weights and stats values (weighted average) but the stats weights are a function of the value themselves. Therefore you can't really use a software solve approach for this because there is a loop. You can't find the ideal amount of a stat if its value is dynamic, at least not using this solver only. I think it's possible with some automated process in a VBA fashion, but this out of my range of competence.

The best you can do is use the table within the xls and do it by hand and trial and error. Input hypothetical stats mixes which total is still equal or very similar to what you have, and try to reach a "score" thats higher than what you current gear provides.

Even then, I seriously doubt of the value of such a mathematical approach to the problem since when I did it, it ALWAYS told me to put all I could in mastery (which makes sence since the weighting is constant AND high). It never pointed towards some "breakpoints" in haste. In reality, for gameplay flexibility among other reasons, these mathematical solutions don't apply since you still want some haste in a way.

I asked for the graph aswell with no success on the discord so I tried my best to "replicate it".

I added points at every 500 increment in haste, and then used the tools within Excel to model a trendline following them as close as possible (6th degree polynomial was the best fit).

Excel remodelling : http://puu.sh/vEazG/f17bfe9b4a.png

Mastery : y = 2E-23x6 - 2E-18x5 + 4E-14x4 - 6E-10x3 + 5E-06x2 - 0,0222x + 40,08 with R² = 0,986

Haste : y = -3E-22x6 + 2E-17x5 - 5E-13x4 + 7E-09x3 - 6E-05x2 + 0,2218x - 366,83

R² = 0,9968

Alternatively instead of taking these to compute the best stat mix given a stat budget, you can do it more simply with a table.

Here is the full xls : http://www28.zippyshare.com/v/YfHegz49/file.html

If anyone can direct me to an uploader that looks less fishy than that, I can totally reupload.

With this ideally you want to use Excel solver (downloadable but not included in it by default as far as I know).

http://puu.sh/vEaoM/6147ff2385.png

Objective is to maximize score by changing variable cells "Haste" and "Mastery".

This looks great and all but in the end all I just said is useless, because of the "loop" phenomena that takes place.

The "score" you want to optimize is a function of stat weights and stats values (weighted average) but the stats weights are a function of the value themselves. Therefore you can't really use a software solve approach for this because there is a loop. You can't find the ideal amount of a stat if its value is dynamic, at least not using this solver only. I think it's possible with some automated process in a VBA fashion, but this out of my range of competence.

The best you can do is use the table within the xls and do it by hand and trial and error. Input hypothetical stats mixes which total is still equal or very similar to what you have, and try to reach a "score" thats higher than what you current gear provides.

Even then, I seriously doubt of the value of such a mathematical approach to the problem since when I did it, it ALWAYS told me to put all I could in mastery (which makes sence since the weighting is constant AND high). It never pointed towards some "breakpoints" in haste. In reality, for gameplay flexibility among other reasons, these mathematical solutions don't apply since you still want some haste in a way.

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **sodcross** » 03 May 2017, 13:56

is there a way to convert pawn stat weights to ask mr robot shadow priest stat weights 7.2 ??

Thx

Thx

### Re: [H2P][7.1.5] Raid Stat Weights and Discussion

by **Moranall** » 10 May 2017, 12:54

I am having an extremely difficult time understanding the suggestions of 10k haste and the 12k haste. I didn't raid in EN so I am not aware of the stats/mechanics from then, so maybe this is old rhetoric that is carrying over?

10k haste - why does this keep being suggested? Per the sims and stat weight charts, mastery is considerably more valuable than haste once you're over 8k haste. In my raiding, I've found that building around mastery has been the best approach for DPS. Why is 10k haste still being the recommended focal point?

12k haste - why does this keep getting called a "cap"? There is no "cap" to shadow priests - at 100% haste, we reach the GCD cap, but that's it - as haste affects our DOTs indefinitely. I've seen it called a "breakpoint" as well, but again, I'm not seeing why? I understand around 12k is when the second haste "hill" starts to drop again, but that's still not a "breakpoint." A breakpoint would be like when you get enough haste to squeeze in an extra ability into a time window - something we're not doing.

Am I missing something here? It seems clear as day that mastery is superior to haste (once haste is over 8k for SL, at least). I feel like there is a lot of bad advice out there of all these suggestions for spriests to get to 10k or 12k haste when I think the advice should be "get haste to 8k then focus on Mastery > Haste >>> crit/int"

10k haste - why does this keep being suggested? Per the sims and stat weight charts, mastery is considerably more valuable than haste once you're over 8k haste. In my raiding, I've found that building around mastery has been the best approach for DPS. Why is 10k haste still being the recommended focal point?

12k haste - why does this keep getting called a "cap"? There is no "cap" to shadow priests - at 100% haste, we reach the GCD cap, but that's it - as haste affects our DOTs indefinitely. I've seen it called a "breakpoint" as well, but again, I'm not seeing why? I understand around 12k is when the second haste "hill" starts to drop again, but that's still not a "breakpoint." A breakpoint would be like when you get enough haste to squeeze in an extra ability into a time window - something we're not doing.

Am I missing something here? It seems clear as day that mastery is superior to haste (once haste is over 8k for SL, at least). I feel like there is a lot of bad advice out there of all these suggestions for spriests to get to 10k or 12k haste when I think the advice should be "get haste to 8k then focus on Mastery > Haste >>> crit/int"

Users browsing this forum: No registered users and 1 guest

- Board index
- All times are UTC - 8 hours [ DST ]