Discipline HPM Stat Weights and Optimisation Spreadsheet

[Keiyra]

OK, here's something I should probably know. If PWS doesn't crit, can it have a crit multistrike?

Yes, multistrikes crit independently from the original attack/heal. Each MS will have its own chance to crit and doesn't care whether or not your PWS already crit.

(so yes, the biggest you can get is PWS Crit, MS crit, MS crit. - Enjoy your rng lottery win

[Deleted]

I was asking more in relation to how myllior calculates the crit + ms portions ^^

[Lexabell]

Excluding critical strikes and/or multistrikes, the formula for PW:S as Discipline is,

PW:S = (1.3)(4.59*Spellpower)(1 + Versatility)(1 + Mastery)

To include the average increase due to critical strikes and/or multistrikes, include factors of (1 + Critical Strike) and (1 + 0.6*Multistrike) respectively.

Are the Mastery and Versatility values % values or rating values? Say I have 51% mastery would I put .51 into the equation or 51? Same for vers.

[Myllior]

OK, here's something I should probably know. If PWS doesn't crit, can it have a crit multistrike?

If you were asking about almost any other spell, the answer would be "yes", but for PW:S and CoW (and also Sacred Shield as far as I know), the answer is "no"; whether or not multistrikes are critical is dependent on whether or not the base hit was critical. That being said, this doesn't have any impact on the average increase due to critical strikes and multistrikes; i.e. it still amounts to including factors of (1+C)(1+0.6*Ms), for Disc.

Are the Mastery and Versatility values % values or rating values? Say I have 51% mastery would I put .51 into the equation or 51? Same for vers.

In spell healing equations like the one for PW:S, wherever Mastery, Versatility, Critical Strike, Multistrike and/or Haste exist, they are decimal values representing the percentage you have in each; i.e. if you had 51% Mastery and 6% Versatility, you would use Mastery = 0.51 and Versatility = 0.06 in the equation.

[Myllior]

Caveat Emptor

Regarding these two issues, which are separating empowered and non-empowered healing of Flash Heal and Prayer of Healing for Discipline and properly separating critical and non-critical healing for Holy, they can now be performed in a few minutes if you have access to MATLAB (many universities allow their students free downloads and home use of MATLAB), as I've written some very crude code that will calculate the numbers without the need for the complicated spreadsheets presented in the opening post under the above heading.

To run this code does still require some manual process, as the analysis requires two .txt files to run. Examples of the various files used, as well as the MATLAB code itself, may be found here. The first file is simply a list of the targets that the code should include in analysing the healing done; at the top of this list should be the number of targets to be considered. Examples of these files may be found in the folder linked above and are named 'Disc_Targets.txt' and 'Holy_Targets.txt'. Currently, the code is written to only track targets with one word names, so it's practically restricted to player characters, as pets and bosses rarely have a single word name in the recorded events on WCL.

The second file is a collection of specific event data; to use the MATLAB code as it stands requires the events of the 'Healing' tab as Holy and the events of both the 'Healing' and 'Buffs' tabs as Discipline (to allow EAA tracking). To extract the events, go to the log you're looking at and select the Priest whose healing you wish to analyse, then go to the relevant tab and select 'Events'. At the bottom of the page, you can then click 'Copy' to copy the entire page of events, which may then be pasted into a text file; this has to be repeated for each page of events, as WCL will only display ~300 at a given time. Examples of files containing the extracted events may be found in the folder linked above; there are two things to note,
1. I've extracted the 'Buffs' tab for Holy in the given examples. This is not necessary to run the MATLAB code for Holy; I did it for testing purposes.
2. The examples for Discipline have the healing and buffs events originating from ALL sources, rather than just the Priest being considered. This is not necessary to run the code for either Discipline or Holy; again, I did it for testing purposes.

With these files extracted, the two for Discipline must be combined into a single file, using the events_combination_imp.m code, which orders everything according to time stamp and also removes duplicate events between the two files. This code does not need to be used for Holy, as only one file needs to be extracted in the first place.

With the two required files now obtained, the relevant code of disc_log_post_analysis.m and holy_log_post_analysis.m may now be run. The three required inputs are the names you gave to your targets files and your events file, and the name of the Priest whose healing you wish to analyse. The output is a set of three matrices showing the effective healing, overhealing and raw healing of each of your spells. These matrices are organised in the same pattern as the spreadsheets of this thread; each row corresponds to a new spell, ordered the same as in the spreadsheets, and the columns are for normal, critical, multistrike and critical multistrike healing.

I don't actually expect anyone to use the code of course; I just thought I'd share it since I was working on something else and stumbled across how to do this

[Argantis]

Firstly, thanks very much to Myllior for the spreadsheets, having a limited understanding how these things work I was able to download and use it reasonably easily. I have entered the details for my character in, and the log file information from our first prog kill (last night) of HM Archi, as I figure this is a good representation of where my priest needs to be at its most effective. The stat weights it has given me are slightly different to that i was using (from the Disc discussion thread on these forums).

My current weights i have been using
Int 1.0
Spell Power 0.90
Mastery 0.82
Haste 0.79
Crit 0.70
Multistrike 0.69
Versatility 0.65

The spreadsheets output is
Int 1.0
Spell Power 0.90
Mastery 0.71
Haste 0.68
Crit 0.66
Multistrike 0.57
Versatility 0.53

Forgive the nooby questions, but it would appear to me that as I am gearing up (714 ilvl) the more valuable int/ilvl becomes, and the less important 'perfect' secondaries are?

also when I get a piece of gear with say, crit multi at a higher ilvl than a piece with say, mastery haste, I 'feel' its not an upgrade, when in fact the stat weights (put into pawn in game) tell me it is.

Is this correct behavior?
thanks again!

PS - Pospospos - I Think we raided together a short while on Frostmourne - (Argantis, Rsham in Siege)

Cheers

[PosPosPos]

Firstly, thanks very much to Myllior for the spreadsheets, having a limited understanding how these things work I was able to download and use it reasonably easily. I have entered the details for my character in, and the log file information from our first prog kill (last night) of HM Archi, as I figure this is a good representation of where my priest needs to be at its most effective. The stat weights it has given me are slightly different to that i was using (from the Disc discussion thread on these forums).

My current weights i have been using
Int 1.0
Spell Power 0.90
Mastery 0.82
Haste 0.79
Crit 0.70
Multistrike 0.69
Versatility 0.65

The spreadsheets output is
Int 1.0
Spell Power 0.90
Mastery 0.71
Haste 0.68
Crit 0.66
Multistrike 0.57
Versatility 0.53

Forgive the nooby questions, but it would appear to me that as I am gearing up (714 ilvl) the more valuable int/ilvl becomes, and the less important 'perfect' secondaries are?

also when I get a piece of gear with say, crit multi at a higher ilvl than a piece with say, mastery haste, I 'feel' its not an upgrade, when in fact the stat weights (put into pawn in game) tell me it is.

Is this correct behavior?
thanks again!

PS - Pospospos - I Think we raided together a short while on Frostmourne - (Argantis, Rsham in Siege)

Cheers

For weapons ilvl is always more valuable(as of this tier), as for the other offpieces that aren't tier pieces, that might not necessarily be true(although it definitely can be), since they don't really scale exponentially like weapons do(6 ilvls on a weapon was essentially a 100++ spellpower upgrade, which trumps any "perfect stat combination" of a ilvl).

As such, the best way is still to take your calculated stat weights for your current gear and manually math out the stat value of your potential upgrades.

Also, hello!

[Argantis]

For weapons ilvl is always more valuable(as of this tier), as for the other offpieces that aren't tier pieces, that might not necessarily be true(although it definitely can be), since they don't really scale exponentially like weapons do(6 ilvls on a weapon was essentially a 100++ spellpower upgrade, which trumps any "perfect stat combination" of a ilvl).

As such, the best way is still to take your calculated stat weights for your current gear and manually math out the stat value of your potential upgrades.

Also, hello!

OK great, yes that makes sense. I shall trust the spreadsheet and the numbers!

Cheers

[Deleted]

-- this post was causing more misunderstanding than enlightment ---

[PosPosPos]

You are also right in your obvservation that int becomes more valueable the higher your itemlevel goes. That is because Intellect always scales linearly with your spells whereas secondary stats scale with some factor ~X^a i.e. a∈(0,1) with strictly positive but diminishing returns. I can't think of a class that does not interact with a sec. stat like this right now.

Eh? I am pretty sure int/sp actually becomes *less* valuable as your ilvl goes up, because as you said, int/sp scales in a "flat" manner and also that it's a lot more accessible than secondary stats. If you noticed, the weights of secondary stats have actually gone up compared to int/sp from HM to HFC.

[Deleted]

Eh? I am pretty sure int/sp actually becomes *less* valuable as your ilvl goes up, because as you said, int/sp scales in a "flat" manner and also that it's a lot more accessible than secondary stats. If you noticed, the weights of secondary stats have actually gone up compared to int/sp from HM to HFC.

Ok, your right. Stand alone my post is plain wrong. I'm talking ceteris paribus, for which my statement is right. Unfortunately I don't have the time to correct/specifiy this right now.

[Myllior]

When looking at things in the formulaic manner that these spreadsheets do, every stat scales as a sum of rational functions (see Stat Weights via Taylor Series), of which the dominant term is always a rational function for which the degree of the numerator polynomial is one less than the degree of the denominator polynomial. (Indeed, each rational function involved in this type of stat scaling has a numerator polynomial whose degree is less than that of the denominator polynomial).

To make this more explicit and also to elucidate why secondary stats creep up in power relative to Intellect and Spell Power as one gears up, consider that basic secondary stats scale like 1/(1+X), whereas Intellect and Spell Power scale like 1/X. Normalising these scaling behaviours with respect to that for Intellect and Spell Power then gives that the normalised weights for secondary stats scale like X/(1+X), which is a monotonically increasing function that approaches unity from below as X approaches positive infinity.

So, as you gear up, not only does the greater availability of Intellect and Spell Power, compared to individual secondary stats, boost the value of secondary stats relative to Intellect and Spell Power, but the different scaling behaviour these stats exhibit also means that, while secondary stats will have an initially lower weight, they will also decrease at a lower rate, hence increasing their value relative to Intellect and Spell Power.

PS: I'm quite tired at the moment, so sorry if I've said something silly in the above.

[PosPosPos]

Ok, your right. Stand alone my post is plain wrong. I'm talking ceteris paribus, for which my statement is right. Unfortunately I don't have the time to correct/specifiy this right now.

Ah I see, then you would have been fine just adding context to your previous post.

[Deleted]

Firstly, thanks very much to Myllior for the spreadsheets, having a limited understanding how these things work I was able to download and use it reasonably easily. I have entered the details for my character in, and the log file information from our first prog kill (last night) of HM Archi, as I figure this is a good representation of where my priest needs to be at its most effective. The stat weights it has given me are slightly different to that i was using (from the Disc discussion thread on these forums).

My current weights i have been using
Int 1.0
Spell Power 0.90
Mastery 0.82
Haste 0.79
Crit 0.70
Multistrike 0.69
Versatility 0.65

The spreadsheets output is
Int 1.0
Spell Power 0.90
Mastery 0.71
Haste 0.68
Crit 0.66
Multistrike 0.57
Versatility 0.53

Forgive the nooby questions, but it would appear to me that as I am gearing up (714 ilvl) the more valuable int/ilvl becomes, and the less important 'perfect' secondaries are?

also when I get a piece of gear with say, crit multi at a higher ilvl than a piece with say, mastery haste, I 'feel' its not an upgrade, when in fact the stat weights (put into pawn in game) tell me it is.

Is this correct behavior?
thanks again!

PS - Pospospos - I Think we raided together a short while on Frostmourne - (Argantis, Rsham in Siege)

Cheers

Those numbers don't look too wrong, it's my understanding that the ones widely posted as stat weights for BiS lists and the like are made using an almost BiS character using some of the top logs on the most healing intensive fights, that's certainly going to change things.

Regarding ilvl vs stats; I'd just trust the stat weights and add them up (presumably that's what pawn does).
For example I have access to 715 crit+mastery bracers or 725 haste and versatility, even if I devalue haste somewhat the 725 still wins for me. On one side you get more int, that's obvious but you do get MORE of the 'bad' secondary also. Often 'more' matters more than 'correct'.

[Argantis]

Those numbers don't look too wrong, it's my understanding that the ones widely posted as stat weights for BiS lists and the like are made using an almost BiS character using some of the top logs on the most healing intensive fights, that's certainly going to change things.

Regarding ilvl vs stats; I'd just trust the stat weights and add them up (presumably that's what pawn does).
For example I have access to 715 crit+mastery bracers or 725 haste and versatility, even if I devalue haste somewhat the 725 still wins for me. On one side you get more int, that's obvious but you do get MORE of the 'bad' secondary also. Often 'more' matters more than 'correct'.

Yeah that is exactly the sort of dilemma I was having, around weapon and headpiece. The way you just explained it makes perfect sense.

In a way its all fairly moot, as me just playing better is probably more beneficial than any of the gear changes, but of course you want to maximize your potential!

[Djriff]

Yeah that is exactly the sort of dilemma I was having, around weapon and headpiece. The way you just explained it makes perfect sense.

In a way its all fairly moot, as me just playing better is probably more beneficial than any of the gear changes, but of course you want to maximize your potential!

Keep in mind, you can calculate your own stat weights using the spreadsheet and your own logs. These stat weights are BiS against top logs, but you may play disc slightly differently than a top tier disc priest (nothing wrong with that), and you'll be surprised how much they differ from fight to fight. Generally speaking, it's safer to just go with the advised stat weights.

[Argantis]

Generally speaking, it's safer to just go with the advised stat weights.

Just to clarify, are you saying its safer to go with the weights supplied on these forums, or your own from the spreadsheet?

thanks

[Djriff]

Just to clarify, are you saying its safer to go with the weights supplied on these forums, or your own from the spreadsheet?

thanks

Well your stat weights will always be more accurate to you, but you would need to average them across several different types of fights. Your stat priority for iron reaver will be vastly different than say HF council or Tyrant. Additionally there are different stat weights when you choose CoW over WoM, but the difference is minute.

I would go with the posted stats, as those are generalized across a lot of different fight types.

[Argantis]

Well your stat weights will always be more accurate to you, but you would need to average them across several different types of fights. Your stat priority for iron reaver will be vastly different than say HF council or Tyrant. Additionally there are different stat weights when you choose CoW over WoM, but the difference is minute.

Ok yep that makes sense. I did just look at archimonde HM, purely as it was the prog. fight we just did, and it was the most demanding of me as a healer so far this tier.
I will reiterate with all bosses next reset and see if there is a difference.

Thanks for the advice!

[Djriff]

Ok yep that makes sense. I did just look at archimonde HM, purely as it was the prog. fight we just did, and it was the most demanding of me as a healer so far this tier.
I will reiterate with all bosses next reset and see if there is a difference.

Thanks for the advice!

Happy to help!