[Resto] Stat weight calculations

[ManyHeadedDruid]

----- Updated for 7.1 -----

Hello fellow theory-crafters,

I was reading about the secondary stat weight situation and found it very disturbing. No one seems to know much and the guides on more prominent sites tend to disagree (AMR, Icy-veins, MMO). I admit healing is very hard to tackle but we do know a few things so I will try to focus on that.

First thing first, here is the healing equation for druid:

T = S ∙ S_c ∙ (1+V/40000) ∙ (1+(4.8%+M/58400) M_s ) ∙ (1+6%+C/35000) ∙ (1+H/32500) + E

where T=Total healing, S=Spell Power (Intellect), S_c is the spell coefficient (the spell specific multiplier) plus traits/talents that increase it, V=Versatility, M=Mastery, M_s=HOT stack count on the target, C=Critical, H=Haste, E=Effects from certain trinket procs.

We also know that certain spells do not benefit from haste (e.g. Tranquility), others crit for 250% (by planting Living Seed) and certain trinket procs do not benefit from anything (Naglfar or Vial). This makes it impossible to give a general answer to the stat weight question without knowing the exact playstyle and content.

Versatility (reliable calculation)

We know that every heal that benefits from spell power also benefits from versatility and vice versa so we can calculate versatility-intellect ratio. It depends on your gear but not on your playstyle or content. It is 0.78 for 32260 Spell power and 1433 Versatility (for any spell, any playstyle, any content). I define versatility value as (how much would I heal more if I had X more versatility)/(how much would I heal more if I had X more spell power).

The general equation is:

S / (40000+V)

Versatility has other effects like increased damage and reduced damage taken but were not included in calculation.

Crit (half reliable calculation) – modified for 7.1

Value of crit is harder to calculate due to 2 mechanics:

• Some spells crit for 250% (e.g. Swiftmend planting seed) so your playstyle changes its value (if you use those spells more often crit is worth more).

• Crit has higher chance to overheal than any other effect which lowers its value by a small amount. This might be insignificant I cannot tell.

It is not hopeless to calculate, however. Let us assume – based on parses – that 10% of total healing plant Living Seed (thus 90% not) and disregard the overheal issue for now. For this I modify the original equation crit part to:

1+(6%+C/35000)∙90% + 1.5∙(6%+C/35000) ∙ 10% = 1.063 + (1.05∙C)/35000

Giving us the crit value:

S / (35333+C)

A few notes about crit:

• Crit is better than versatility however you can overdo it ending up crit being weaker than versatility. The breaking point where versatility gets stronger than crit is 4667 more crit than versatility which is easy to achieve if you ignore all items with versatility.

• Crit heals are the most likely to overheal which reduces its value.

• Crit increases damage output more than versatility if you happen to dps a bit.

Haste (mainly guessing)

Haste would be easier to calculate if all spells would benefit from it but it is not the case. Tranquility – which gives 10-25% of our total healing – does not. Neither Swiftmend and to some extend Regrowth. Checking logs gives the impression that only 60% of our healing is directly amplified by haste but it is not that simple due to our mastery mechanic. Mastery increases almost all healing we do if there is a hot on the target. Even Tranquility is increased if Rejuvenation is running. To capitalize on this mechanic druids can and should amplify tranquility and all other spells with placing rejuvenations/WG/SotF/etc. on targets or when our main combo is executed (Swiftmend+WG+Flourish+Essence).

The goal of haste is not simply the healing output but also to lower GCD to enable placing more hots before Tranq or combo. This has small value but not insignificant. Sadly this value is not expressible by equations so I will do the haste equation without it. Keep in mind that haste is worth more than this.

Calculating with 60% total healing affected by haste gives us the value:

S / (54167+H)

Mastery (reliable if hot count is known, meaningless otherwise)

Mastery is tricky as we need to know the hot count on the target to calculate its value. Placing a hot is already counts as one so a hot itself is increased by mastery. The equation is:

S∙M_s / (58400 +M_s ∙ (2803.2+M))

Of course M_s is the main question here. What we know about hot count:
60-80% of our healing is hot and other spells (like tranq) will hit targets with hots occasionally. We can safely assume that hot count is at least 1 on average.

Spreading hot playstyle cannot hope for much higher but focused healing (especially in small groups) can reach 2+ on average.
At M_s=1 the value of mastery is terribly low: 2/3 of versatility. In a small group, however, mastery starts to shine and becomes the best secondary stat while at M_s=2.18 mastery equals intellect in value (at 32260 intellect and 2707 mastery).

Here is a tool to analyze your log. It shows your weighted average hot count and calculates Mastery stat weight if you enabled Advanced Logging in Network settings.

http://proudguild.eu/resto-stats/

Example

Example restoration druid and the corresponding stat weights:
Intellect: 32260, Versatility: 1433, Mastery: 2707, Critical: 10373, Haste: 5274

Stat weights (intellect=1):

• Versatility: 0.78

• Critical: 0.71

• Haste: 0.54

• Mastery (mythic raid healing): 0.50

Please post if you have any suggestions – especially to the mastery – or point out mistakes. I’m curious to see what you guys think!

[Deleted]

Check out this: https://www.dropbox.com/s/k3sjc4yedh...v1.1.xlsx?dl=0

Click "Calcs" on the lower bottom and this could probably help you with your own stat weight calculation.

[Deleted]

Check out this: https://www.dropbox.com/s/k3sjc4yedh...v1.1.xlsx?dl=0

Click "Calcs" on the lower bottom and this could probably help you with your own stat weight calculation.

Link doesn't work for me :/ Page just ends up blank.

EDIT:
Nvm, opened it in Incognito and got the download.

[ManyHeadedDruid]

Check out this: ...

Click "Calcs" on the lower bottom and this could probably help you with your own stat weight calculation.

Nice work! It gives very similar results with same input stats. Is it based on simulation? Is there a description somewhere how it works?

[Grayclaw1]

Hmmm, I was always under the impression that Haste is our best secondary stat by far. This seems to contradict that notion and further muddy the waters for me regarding what our stat prios should be as a resto druid. Grrrrrrrrrrrr

[ManyHeadedDruid]

Hmmm, I was always under the impression that Haste is our best secondary stat by far. This seems to contradict that notion and further muddy the waters for me regarding what our stat prios should be as a resto druid. Grrrrrrrrrrrr

Don't drop your haste gear just yet! This calculation doesn't (cannot) show the combo's dependency on haste. We Tranq/combo when it is your turn to save the raid. If you simply push Tranq then it will heal medium amount. To heal decent amount we need to set it up properly. Need to spread hots all over the raid then Tranq (or combo) so it heals up to 10%+ more (ppl will have hots from WG, Rejuv, Spring Blossoms, etc.). If we can get an Innervate from our Balance buddies (20% haste), then this goes up higher as u can push out more hots in the time window (Tranq is 8 secs so you have exactly 10 secs to set up as Rejuv is 18 secs, exactly Innervate duration)
I am not sure how significant this is (I guess haste is not as high as crit). But maybe someone will throw in a nice equation here

[Twitchys]

This calculation ignores the direct impact of haste on cast speed/gcd and mana consumption, so it will misstate the value of haste for total HPS. Just a warning before people go burning their haste gear.

The reason haste always showed so high in hamlet's sheets before legion and the reason it's still considered a top stat is because it has the best direct scale factor for raw effect (325), affects all non-cd spells via faster cast rate, and affects all hots via extra tics. This leads to rejuv (and technically regrowth) double dipping (higher healing per cast, and more casts), and at least linear scaling on all of our other spells except SM, our only CD based direct heal. The secondary effect here of increased mana cost, which is where calculating stat value for healers actually gets tough, is also not taken into account which will probably suppress the value of haste based on fight length, but is mitigated by good play.

Additionally, haste increases the proc rate of (some) RPPM trinkets, and could have some increased value due to this. I'm not sure what procs specifically are scaled this way, but I think Vial of Nightmare Fog at least is RPPM and should be scaled by haste.

Haste also should be (not 100% sure here) the best scaling stat for DPS as well (at least caster for dps, feral affinity is probably different), which isn't that important, but probably should influence edge decisions, especially for small group content.

That said, the rest of the calculations seem about right to me, with reasonable results. Overall, I'm going to stick with my strategy of preferring haste, with all of the other stats pretty close behind as a good balance between spread and tank/5man healing mastery value.

Edit: I just wanted to add an edit to thank you for putting forward theorycrafting. With Hamlet gone, we need more people to come forward and open discussions about resto theorycrafting, so kudos to you for doing so!

[Torty]

This calculation ignores the direct impact of haste on cast speed/gcd and mana consumption, so it will misstate the value of haste for total HPS. Just a warning before people go burning their haste gear.

No, it doesn't ignore it. It has Haste HPCT value right there. I think it's impossible to sim Haste and Mastery, so I put multiple values for these stats in the spreadsheet so people can decide for themselves.

Haste also should be (not 100% sure here) the best scaling stat for DPS as well (at least caster for dps, feral affinity is probably different), which isn't that important, but probably should influence edge decisions, especially for small group content.

That needs to be simmed. I believe Crit and Vers are better dps stats if you do a lot of cat dps. If you only wrath, moonfire and sunfire, then Haste has no competition, probably even better than Int dps wise then.

That said, the rest of the calculations seem about right to me, with reasonable results. Overall, I'm going to stick with my strategy of preferring haste, with all of the other stats pretty close behind as a good balance between spread and tank/5man healing mastery value.

Edit: I just wanted to add an edit to thank you for putting forward theorycrafting. With Hamlet gone, we need more people to come forward and open discussions about resto theorycrafting, so kudos to you for doing so!

No problems. Just use the spreadsheet as a guidance, not stat rule. You can see how each spell scales with different stats, you can see what gives you more healing per mana and what gives you more healing when you want to spam. Use the stat values accordingly.

- - - Updated - - -

Hmmm, I was always under the impression that Haste is our best secondary stat by far. This seems to contradict that notion and further muddy the waters for me regarding what our stat prios should be as a resto druid. Grrrrrrrrrrrr

It neither contradicts that notion nor proves it. You get 2 values for Haste. One is how much healing it adds per mana spent (Haste HPM) and another is how much healing it gives when you spam cast without caring about mana (Haste HPCT). Both are valueable numbers and you need to decide for yourself what you want to get from Haste. Maybe some middle ground or only HPM or only HPCT depending on situation.

For example, in dungeons you should definitely use HPCT value since you don't care about mana there and only about how fast you can heal people up. That's obviously not the case in raids.

[Twitchys]

@Torty I haven't looked at the sheet. Was responding to the top of the page. Based on your response it sounds like it's a lot like Hamlet's old sheet and it also sounds like you understand what you're talking about, so I'll have to take a look when I get home and can get on dropbox.

[Tomali]

This is a really cool and interesting post!

The value of haste and our 'double dip' that people talk about would effectively change the spell coefficient (Sc) due to adding ticks to our HoTs, which I think would raise your calculated haste value for lifebloom, CW, rejuv, WG, and RG.

This calculation would work for swiftmend, healing touch, and other direct heals, but adjusting the spell coefficient due to haste is likely what makes the stat so strong.

[ManyHeadedDruid]

Additionally, haste increases the proc rate of (some) RPPM trinkets, and could have some increased value due to this. I'm not sure what procs specifically are scaled this way, but I think Vial of Nightmare Fog at least is RPPM and should be scaled by haste.

This is interesting, is it true, anyone tested it? This could be inserted into the equation.

And haste scaling is indeed the best (as in the equation) but its usefulness is limited as certain spells are not affected. I do not say haste is bad though. We need to think more about it

- - - Updated - - -

It neither contradicts that notion nor proves it. You get 2 values for Haste. One is how much healing it adds per mana spent (Haste HPM) and another is how much healing it gives when you spam cast without caring about mana (Haste HPCT). Both are valueable numbers and you need to decide for yourself what you want to get from Haste. Maybe some middle ground or only HPM or only HPCT depending on situation.

For example, in dungeons you should definitely use HPCT value since you don't care about mana there and only about how fast you can heal people up. That's obviously not the case in raids.

I agree.
This calculation is based on logs, these values have nothing to do with mana or maximum possible HPS. This is how i ended up with 60% of spells are affected by haste. The spreadsheet is another approach where you can decide what you want to do (for example go for high HPS). Here it is already decided what happened (the log tells) and if you repeat the encounter in a similar way then the best would be to increase your selected stats. Both approach can be useful. I think the spreadsheet is good for planning and this is good after the fight how to optimize/progress. Maybe.
Beside this: if you have high haste then you need to use such spells to not waste it.

Thanks for the feedback, guys, i really appreciate it!

[Cheze]

you can't really evaluate mastery without considering talents and fight circumstances. My experience (using the RDSW WA linked earlier) is that on any fight where cultivation sees significant uptime mastery pulls ahead of versatility (to approximately 75% the value of crit on my gear.) Spring blossoms likely has a similar (if smaller) impact, although most druids seem not to use it on many fights.

Inner Peace also tends to shift your weights in favor of crit, since (other than vers) it's the only stat that meaningfully increases the value of tranquility. RDSW generally reports that crit is my highest value rating even though it's also my highest rating in raw terms (though by less of a margin than in OP's example.)

[Alltat]

No, it doesn't ignore it. It has Haste HPCT value right there. I think it's impossible to sim Haste and Mastery, so I put multiple values for these stats in the spreadsheet so people can decide for themselves.

The top post doesn't mention HPCT in the haste entry, and instead claims that only 60% of your healing benefits from haste, which isn't true. All of your spells (but not all procs) benefit from haste, and then in addition to that, 60% of your healing double-dips and benefits from haste twice. For Rejuvenation you get both a faster casting time (lower GCD) and more ticks. For Swiftmend you only get the GCD reduction, but that's still a HPS gain. Being able to cast Rejuvenation faster also means that some things scale indirectly with haste, notably Dreamwalker and Cultivation.

The only things that don't scale with haste at all are Nature's Essence, Ysera's Gift, and the Lifebloom bloom.

[ManyHeadedDruid]

you can't really evaluate mastery without considering talents and fight circumstances. My experience (using the RDSW WA linked earlier) is that on any fight where cultivation sees significant uptime mastery pulls ahead of versatility (to approximately 75% the value of crit on my gear.) Spring blossoms likely has a similar (if smaller) impact, although most druids seem not to use it on many fights.

Inner Peace also tends to shift your weights in favor of crit, since (other than vers) it's the only stat that meaningfully increases the value of tranquility. RDSW generally reports that crit is my highest value rating even though it's also my highest rating in raw terms (though by less of a margin than in OP's example.)

I agree. Mastery is very hard to judge, that 60% might be true for certain fights/builds and false for others. I have not checked logs where Cultivation was used though(anyone uses it for raiding?), i strictly checked druids with SotF. Small group healing picks Cultivation and then Mastery is the king. While I have no idea how weak Mastery is in raid healing and how strong it is for small groups, I assumed it is "0.5" weak for raids and "1+" strong for small groups. I have no calculation to support this. Is it possible to know how many hots did I have on a boss fight from logs? Can anyone help out?

And I agree with Inner Peace comment as well. If it is possible to use Tranq more then picking up Inner Peace and actually using Tranq will increase crit's value.

- - - Updated - - -

The top post doesn't mention HPCT in the haste entry, and instead claims that only 60% of your healing benefits from haste, which isn't true. All of your spells (but not all procs) benefit from haste, and then in addition to that, 60% of your healing double-dips and benefits from haste twice. For Rejuvenation you get both a faster casting time (lower GCD) and more ticks. For Swiftmend you only get the GCD reduction, but that's still a HPS gain. Being able to cast Rejuvenation faster also means that some things scale indirectly with haste, notably Dreamwalker and Cultivation.

The only things that don't scale with haste at all are Nature's Essence, Ysera's Gift, and the Lifebloom bloom.

The calculation is based on logs and does not (cannot) assume that you would have cast more spells if haste was higher (double dipping is only meaningful in evaluating potential). The calculation assumes that you spent all your mana and thus could not have cast any more spells. I know this is false but i don't know by how much. Is it significant error or can be ignored? This results in haste value being underestimated (a fact) by an unknown amount (small? amount). i would appreciate if someone tried to put a number to this with some proof.

By the way you can remove Ysera's Gift from your list as it doesn't scale with Intellect either (like trinket procs).

[Cheze]

I use cultivation on ursoc, dragons, ilgynoth (though I'm not settled that it's the right choice there) and cenarius. It might even be worth using on nythendra but I haven't ever tried it, and on elerethe the damage is mostly trivial either way.

I don't think you can pull any kind of an accurate mastery contribution out of log data; you'd have to go back through each individual's buff uptimes and track the HoT overlap and calculate the increase in healing done, and I don't there's any way to make a simple filter that does that. I guess you could look at actual average HoT ticks vs. your base HoT ticks to try and get an idea of how much bonus healing mastery is generating, but that would be an extremely rough estimate and I've never bothered with it.

[Keiyra]

• Some spells double crit (e.g. Swiftmend planting seed) so your playstyle changes its value (if you use those spells more often crit is worth more).

What do you mean double crit? Living seed is not another 100% on top of the heal. Seed is only an additional 25% (or 49% with 3/3 ranks) - and it almost only has any value when applied to tanks. - Its unclear from your description, but you really shouldn't add a ton of a value to crit due to Living Seed. It's a rather minor effect (especially now in 7.1)

[ManyHeadedDruid]

What do you mean double crit? Living seed is not another 100% on top of the heal. Seed is only an additional 25% (or 49% with 3/3 ranks) - and it almost only has any value when applied to tanks. - Its unclear from your description, but you really shouldn't add a ton of a value to crit due to Living Seed. It's a rather minor effect (especially now in 7.1)

I wrote it before 7.1 without that trait (I assumed we don't have it yet). Living Seed was 50% of the total heal which made that heal 300% (200% is the crit and +50% of that - i called it double crit).
Now it is 250% and I will update original post to reflect that. It will lower crit's value.
Beside this you are absolutely right, Living Seed isn't a major effect and I suspect it increases crit's value by little, i haven't done the calculation without that however, so cannot tell.

[Drekor]

I'm a little surprised at just how well Mastery scales, 2.18 to be equal to int? In high tank damage scenarios I have 6 running. In high group damage scenarios I have 4 running on everyone. I knew it was my best stat for mythic dungeons but according to these numbers I should be dropping my int gem for a mastery one @_@

[Alltat]

The calculation assumes that you spent all your mana and thus could not have cast any more spells. I know this is false but i don't know by how much.

By a lot. People usually have a fair amount of mana left at the end of a fight, or at least enough to cast a dozen Rejuvenations. With spirit not being a thing anymore, haste is also the only stat that increases how much mana we have. Faster Lifebloom ticks means more Clearcasting procs, and faster casts means you can cast more spells during Innervate.

I think you might get better results if you go by logs and assume that mana is infinite. Assume that any time where the druid didn't cast anything was time where no healing was needed, and that any time the druid spent casting was time when more healing would have been useful; this is usually true. Then just see how much "rejuvenation spam time" is freed up by a point of haste.

Synergies make this a bit messier, but you can probably assume that these extra Rejuvenations contribute just as much to Cultivation and Dreamwalker healing as other rejuvs.

[Torty]

Assuming mana is infinite and time spent not casting was when healing wasn't needed is one of the dumbest things I've read in a while. That's not how healing works. If I could spam the shit out of my spells every time healing was needed, I'd be playing completely differently.