Resto Druid: Ask Mr. Robot Stat Weights Explained

[Swol]

Hey all, I'm Swol, the main theorycrafter at Ask Mr. Robot and the person responsible for most of the rotations, stat weights, etc. on the site.

I've made little posts here and there regarding our resto druid weights. This will be a compilation of a lot of the reasoning that goes into healing stat weights in general, as well as resto druid in particular.

Part 1: Max Theoretical Healing vs. Topping the Meters
Part 2: Haste and Mastery for Resto Druids
Part 3: Why not stack HoTs in raids?

Part 1: Max Theoretical Healing vs. Topping the Meters

Healing is a fundamentally different problem to solve than DPS. For DPS, there is no limit to how much damage per second you can do. The fight will simply end sooner if you do more damage. For healing, there is a limit to the HPS you can do: the damage the raid takes. I will start with an example I like to give, which illustrates why healing meters can be very, very misleading for healing theorycraft:

Assume your raid takes 400,000 damage per second, on average. Now, assume you have three identically geared and skilled healers each capable of putting out 150,000 healing per second, on average. Their potential is 450,000 healing per second. But... you don't NEED that much healing. It is very rare that those three players will all be able to do 133,333 HPS on the fight. Assuming they are all playing different specs, chances are that one of those healers will be good at snagging small amounts of effective healing the fastest, allowing them to do 150,000 HPS. So, now the other two healers are left doing 125k each, making them look like they aren't playing as well. It is very rare that you need exactly 3 or exactly 4 healers on a fight. You usually need 2.75 healers, or 3.5 healers, etc. This is why healing meters and especially healing log rankings are almost completely worthless for measuring performance or theoretical stat analysis. Topping meters is a fun meta game, but, we have to admit that it is just that: a meta game. You can be a perfectly effective, skilled, and useful healer without topping a meter.

This leads to a situation where healers are competing with each other to grab the effective healing first, so that they look the best on the meters. Another consequence of this situation is that people will start to gear in such a way that they are able to be the first one to heal people. The stats that could get you the most healing possible over the course of a fight might not be the best stats for being the FIRST one to heal people on fights. Resto Druids are an extreme example of this in Legion.

Part 2: Haste and Mastery for Resto Druids

Haste is a tough stat to deal with in healing theorycraft. Its value depends heavily on how long a fight is. On longer fights, you are limited by mana, not time, thus making the ability to cast more spells pretty weak. But, resto druids heal almost exclusively with HoTs. As we know, HoTs actually get a throughput increase from Haste independent of the quantity of HoTs you cast. Each HoT will actually do more healing over its fixed duration.

What else does Haste do for us? It helps us top meters. More haste on all those HoTs that are all over the raid also translates to increased heal frequency. All of those HoTs are excellent at snagging up small bits of incidental damage FIRST. This makes you look great on those healing meters.

Mastery for resto is new in Legion. You do more healing per HoT on your target. This stat offers the potential for extremely high theoretical healing throughput. Mathematically, it is the superior stat for resto druids. But, of course, it depends on you playing in such a way that you are stacking HoTs on targets instead of spreading HoTs around.

This leads us to the root of why "they" recommend Mastery for 5 mans and Haste for raids. In a 5 man, you don't have anyone competing with you for the available healing. It's up to you to do everything, giving you maximum opportunity to stack HoTs. In raids, most of the time we are more or less playing wack-a-mole with our raid frames as fast as we can. Putting up HoTs on whoever, wherever - actually fighting the other healers for heals a lot of the time.

Here are a couple of simulations that quantify the difference in healing you do when you stack hots in a 20 person raid, vs spreading them out:
A couple of notes regarding the simulations PLEASE READ THESE NOTES BEFORE LOOKING AT THE SIMULATION RESULTS:
1.) I used a 34 point artifact just to avoid biasing results due to the artifact path chosen, when you do your own simulations on your character, results can vary somewhat, but none of the artifact traits make a drastic difference in stat priorities. This also partially explains the very high HPS numbers you will see, though.
2.) This is assuming that all healing you do can be effective healing and that there is always someone to heal. There is no overhealing. Why? Because, I want to find out how much healing is possible, and which stats are best for that. If the raid doesn't need all of my healing, then I'm not going to be as concerned about what stats I use. In that case maybe I would just gear for topping meters.
3.) If you are doing your own simulations, note that cultivation is not yet implemented in the simulator. I hope to have health-based effects like that available next week sometime.
4.) I just picked some common talents. I am not in any way saying the talents chosen are necessarily the best talents.
5.) I didn't use any consumables and the rotation doesn't have bloodlust in it right now.
6.) The fight length is 6 minutes +/- 20% to approximate a fight that isn't crazy long, but not really short either.
7.) The "Haste/Crit" gear set I created by taking 8000 rating out of mastery and splitting it between crit and haste. I didn't change the actual gear, I just did a stat override to keep it simple.

There are two rotations I'm using for tests:
Max Healing - this rotation will stack HoTs on the tank and off tank, it puts efflorescence in the group of ranged players and uses wild growth on those players. AMR weights were calculated based on this healing pattern which leverages resto's mastery.

Spread Healing - this rotation will spread HoTs around the raid evenly and alternate using wild growth on the melee and ranged groups. It looks for players with no HoTs on them to heal first.

Simulation 1: Max Healing 840 Mastery-heavy gear set (440k HPS):
http://www.askmrrobot.com/wow/simula...8d4710a838b569

Simulation 2: Max Healing 840 Haste/Crit gear set (426.5k HPS):
http://www.askmrrobot.com/wow/simula...2408b9fafa9961

Simulation 3: Spread Healing 840 Mastery-heavy gear set (349k HPS):
http://www.askmrrobot.com/wow/simula...150680baa93749

Simulation 4: Spread Healing 840 Haste/Crit gear set (361.5k HPS):
http://www.askmrrobot.com/wow/simula...942f409acf6a08

So, as you can see: If you don't try to stack HoTs on targets, Mastery will be worse than Haste/Crit. If you do stack HoTs on targets, Mastery will be better. AND, you will do a LOT more healing if you stack HoTs on targets, regardless of whether or not you specifically gear for Mastery. This leads us to Part 3.

Part 3: Why not stack HoTs in raids?

We see now that when you don't stack your HoTs, you leave a LOT of healing on the table. Why would you do that? Why would your raid team do that? The amount of mana you have available in Legion is never going to go up significantly. We will be mana-limited on the harder, longer fights for the entire expansion. I think it is time we do away with the WoD-style spam heals everywhere mentality. If we look at the mastery of many of the healing specs, it requires some coordination to get the most potential healing out of each healer. Getting the max potential healing out of each healer makes the game more fun for the healers, as well as lets your team start to push towards doing content with one less healer.

Druids: Need to stack HoTs on targets.
Paladins: Need to be close to targets.
Shamans: Need to prioritize lower health targets.
Monks: Need to combo essence font and other heals on targets.
Disc: Needs to heal/shield first, then AoE heal through damage
Holy: Only one that doesn't really need a special play style to benefit from mastery.

I'm lobbying for people to take a look at their healing team and leverage their strengths instead of just "healing whoever". If you have a druid, why not tell your other healers to lay off putting HoTs on, say, the melee players and let the druid stack hots on them. The other healers can use their hots on the ranged players or tanks. Or, you could have the druid keep tons of hots on both tanks and tell other people to lay off them. Why not stand your paladin next to some people and make them more responsible for those players? It takes a little more setup, but, for people running with a regular raid group, even some minor coordination could result in significantly more effective healing.

[RestoSpirit]

Hey,

Im glad there is finally someone who explains the "weird" stat-weighs on AskMr.Robot so first of all thank you for your work a really well written explanation. But I personally disagree with the conclusion.
As you said yourself these calc's are based on the fact that you don't overheal, your theoretical maximum output, but you almost never need this. Druid-mastery becomes much less viable when you produce overhealing.
In a realworld scenario you rather heal 5 people for 10% of their HP then 1 person for 50%, that's what makes spreading hots all over the raid way more viable, which means mastery becomes less useful. Your job as a healer is to keep the raid alive and it's much more likely that someones dies when they don't get heal at all compared to when they at least get 10% from your hot. So in my example we would put 4 people in danger to top of the one you stack your hots on.

Another problem is that you need a lot more Masteryrating to reach 1 % mastery compared to Hasterating needed for 1% haste, which makes it the more effective stat.

So whats wrong with haste? Your mana. You work through your mana in about 2-3min if you blanket the raid with reju and use WG on CD but Legion gives you a great tool to counter that: Innervate. Especially Rapid-Innervation from Balance-Druids is extremely powerful. So when we reduce a healer we are more likely to sub-in a Balance-druid to ensure to have enough mana. I think that's a major problem with simcrafting healers, in the realworld we have a raidsetup, which means we have things like multiple innervates to consider when we choose our gear.

Cheers
RestoSpirit

[Keiyra]

I have to agree with restospirit. I applaud the work you've done, but you cant just ignore overhealing and push towards maximum potential output.

In a raid whenever you stack multiple hots on a target, you essentially devalue each hot as the likelihood of overhealing increases - with perhaps the except of tanks. In addition, as a hot healer, you are trying to stabilize as much of the raid as possible such that no one dips too lows and dies. In terms of preventing deaths, its far better to spread your hots around to as many targets as you can, versus stacking them up on only a handful.

So while its true you CAN put out more potential healing with mastery and stacking (even though overheal will cut into much of that theoretical gain)...you are doing so on far far fewer targets, diminishing your strength as a raid stabilizer, and increasing the risk of raid member deaths.

[RestoSpirit]

Another interesting fact I noticed while going over my logs to compare them to your simulations is the amount of reju casts.

https://www.warcraftlogs.com/reports...aling&source=8

While my fight is 40sec longer, the amount of WG casts is the same as your simulation (haste/crit maxheal) meanwhile I almost triple your reju casts. Looking at the Crit/haste-spreadhealing your reju casts are even lower while WG goes up by 50%. Doesn't make sense to me, since the amount of reju casts should go up when you spread your healing on the entire raid, right?

Cheers
RestoSpirit

[Swol]

I guess I was arguing for more coordination so that healing 2 people for 50% of their HP is better than healing 5 for 20%, since the other healers are all doing the same as well, and you aren't overlapping. You could theoretically all do more healing for less mana. We're so conditioned to spread heals around evenly, that we're leaving so much healing on the table. I think it's a waste... and really not that fun compared to what is possible.

I am trying to show that we don't HAVE to take the position that it isn't viable to stack HoTs in raids. It could be if we tried, and then the healing potential goes up significantly. It's certainly easier to just not worry about it, but, I don't mind putting in the effort to make healing a bit more interesting.

[Keiyra]

The more coordination angle only goes so far - particularly with the semi-smart and aoe heals out there. The possibility for other healers stuff bouncing around and healing your "hotted" targets immediately begins to lower your output and theirs by "breaking coordination" and producing overheal.

In a world of perfect coordination and healing assignments, with 100% control of who our heals will heal...sure, what your proposing would be better, and potentially more fun. In the current incarnation of WoW...I don't know if the efficiency and output gains would materialize as well as you would hope.

[RestoSpirit]

I guess I was arguing for more coordination so that healing 2 people for 50% of their HP is better than healing 5 for 20%, since the other healers are all doing the same as well, and you aren't overlapping.

But wouldn't crit better in that case, since it would make it possible to reduce the amount of healers while still blanketing the raid and having the extra healing from crit to make up for the lack of the additional healer.

For me personally the most fun experience is to heal entire encounters on my own, reducing the amount of healers to just one with everyone avoiding damage as good as possible and playing the druid to it's limits, in that case I'm forced to be GCD-capped to heal everyone and again mastery becomes less viable.

But I get your point, you are tired of spamming reju the whole night and you are looking for a different approach.
Is it possible to play mastery sucessful in a raiding-scenario? Yes, of course. Is it the most optimal way during progress (or later when you reduce healers during farm)? Probably not, since druid excels at keeping everyone alive "a little bit" while other healers top them of.

With more gear mastery is even worse, since you get more int which boosts your overall healing a lot and you run into overhealing again. :/

Cheers
RestoSpirit

[Swol]

Well, I'm not assuming that you can control all the healing. There will always be uncontrolled AoE splash. I was talking about keeping some more control over where people place their targeted heals/hots. You'll never reach the theoretical max unless you solo-heal, but, you can get closer to it than we are. I know it's a long shot because it would require a lot of effort, but, hey, it's still fun to find out what is actually possible.

Either way, the "Spread Healing" rotation is available in the simulator now for people to play around with. I think it will match up with what a lot of people are doing in raids, so you can try things out in the simulator to find stat weights, etc.

We have more healing updates coming soon. I've added a "try random ally" ability to the rotations as well. So, instead of actively looking for someone with no HoTs, it will pick a target totally randomly. This will result in sometimes stacking HoTs, and sometimes not. Could be interesting for gearing theory.

[RestoSpirit]

Either way, the "Spread Healing" rotation is available in the simulator now for people to play around with. I think it will match up with what a lot of people are doing in raids, so you can try things out in the simulator to find stat weights, etc.

We have more healing updates coming soon. I've added a "try random ally" ability to the rotations as well. So, instead of actively looking for someone with no HoTs, it will pick a target totally randomly. This will result in sometimes stacking HoTs, and sometimes not. Could be interesting for gearing theory.

Sounds really cool, is there a way to configure the rotation that it prioritizes reju blanketing over using OoC or LB for example? And is it possible to add how many innervates you will get? Guess not everyone has 4 balance druids in the raid, but it would be cool to optimize your gear that it fits your raidsetup. In a 7min fight Im able to cast 1:30min for free, guess that is a big factor concerning the theoretical output.

Cheers
RestoSpirit

[Cheze]

I don't think the 'why not stacking hots' discussion is very practical in real encounter terms, for a variety of reasons

1) there are now more smart heals flying around the raid than there were in WoD. Resto shamans, hpriests, monks and even other rdruids do not necessarily have fine degree of control over who they're healing, aside from that their targets will be ones taking damage. It is fine in theory to say you can make a resto druid responsible for melee or group three or whatever; in practice, chain heal does what it wants.

2) 'stacking hots' on multiple targets is a strategy that takes time that you often do not have. Spending several globals to stack hots on 3-4 targets (say, swiftmend > WG > rejuv > rejuv > rejuv) produces a high amount of theoretical output but is not particularly helpful if group members need healing faster than that.

3) stacking hots on anything other than tanks is not really very practical; few 20+ raid builds use germination and regrowth is not really sustainable outside of clearcasting procs, so the only real way to deliberately leverage mastery for raid healing is casting WG and chasing it with rejuvs. Which isn't an awful strategy in the face of high incoming damage, but since we don't control WG it's hard to do it in a way that won't produce substantial overheal unless the raid faces overwhelming damage (mythic ursoc.)

in actual practice our preferences will depend heavily on fight/talent choices. Mastery obviously benefits greatly from cultivation (ursoc, cenarius, ?) and also from fights where the raid is split into smaller groups and your healing is naturally focused on fewer targets (in EN, basically just dragons.) We also see from your sims that the performance difference for the crit/haste build is smaller, which may be relevant for gearing choice while we're still at the point in the life of the raid that gear is a limited resource.

[Sprucelee]

Designing a model (and then basing stat weights) with the assumption that players will adapt to it, rather than just modeling how people actually play as a team, after a decade of refinement (i.e. the general rejection of strict assignment based healing in raids) in inherently flawed. Trying to reconstruct the dynamics of how healing teams work to that degree, in order to validate your results, just seems a little naive. It is no secret that mastery COULD be powerful if everyone played around you, and you played specifically to its strengths, but when it comes to actually progressing, that just does not work in a team, there naturally has to be overlap for everyone to be at their strongest potential simultaneously.

You should take some time and play with the RDSW Weakaura across different formats to see how they perform in a realistic environment where you are part of a team and not molding everything to your advantage.

[Swol]

Sounds really cool, is there a way to configure the rotation that it prioritizes reju blanketing over using OoC or LB for example? And is it possible to add how many innervates you will get? Guess not everyone has 4 balance druids in the raid, but it would be cool to optimize your gear that it fits your raidsetup. In a 7min fight Im able to cast 1:30min for free, guess that is a big factor concerning the theoretical output.

Cheers
RestoSpirit

I don't have innervates in the simulator yet - but I could add it. Could be tricky to pick how many, but, would be easy to just make it a spell that the healer could use on themselves, and then you could edit the rotation to use it when you want. And then just modify the cooldown on the spell to take into account how often you want to use it.

The rotation the healer uses is totally configurable. You can take one of my default rotations and copy it, then make it heal however you would like. Hit me up in our discord or forums if you need help with editing them - the healer rotations are a bit tougher to modify than the DPS rotations because of all the ally targeting.

I would definitely encourage people to modify the rotations to heal how they want to heal, and then build gearing strategies around that. If you set the sim type to "gearing strategy" it will create stat weights and trinket/legendary/set rankings for the setup you specify. You can save that and use it with the AMR site to rank all of your gear. Be warned that resto druids are, unfortunately, the slowest spec in the game to simulation, so, gearing strategy runs will easily take an hour or more, even with high margin of error. We've actually improved the speed a lot, but, all those HoTs!
@Sprucelee I'm not saying that everyone has to agree with me or change what they are doing. I just wanted to share with folks that I have well thought out reasoning behind my first set of weights. They will be just one of many sets of weights available by default on the site, we're in the process of making more rotations and doing more calculations. I personally think it is interesting to know the maximum theoretical output and what gear would be best for that. Whether or not people can or want to do what is necessary to achieve that is up to them, but, it is an objective measure of the potential of a spec. I personally think the default rotation could be achieved easily in a raid with one resto druid and some other healers. Just let the resto druid stack HoTs on the two tanks (and take germination), and tell the other healers to only help in emergencies. That alone would be enough to make mastery just as good a stat as haste and crit, and would take almost no coordination.

[Keiyra]

Just let the resto druid stack HoTs on the two tanks (and take germination), and tell the other healers to only help in emergencies. That alone would be enough to make mastery just as good a stat as haste and crit, and would take almost no coordination.

So, basically you'd have to raid without a holy pally or a disc priest. Or raid with a hpally/disc priest who don't mind gimping their effectiveness to boost yours.

[Yellowfive]

@Keiyra Or as a holy paladin, you either have to raid without a resto druid, or raid with a resto druid who doesn't mind gimping their effectiveness to boost yours.

[Keiyra]

@Keiyra Or as a holy paladin, you either have to raid without a resto druid, or raid with a resto druid who doesn't mind gimping their effectiveness to boost yours.

Firstly, your comment doesn't make sense regarding the topic at hand, which was mastery focused druid build, and comes across as an unnecessarily salty comment by a hpally who didn't read the whole thread (that may not be the case, but that's what it seems like)

Secondly, the druid can just go haste/crit and heal the raid as normal without hurting the pally by expecting him to not use his beacon on the tank - which is what the prior suggestion (of going heavy mastery and focus on tanks and asking the other healers to not heal them except in emergencies) would've required.

[RestoSpirit]

The rotation the healer uses is totally configurable. You can take one of my default rotations and copy it, then make it heal however you would like. Hit me up in our discord or forums if you need help with editing them - the healer rotations are a bit tougher to modify than the DPS rotations because of all the ally targeting.

I already tried to modify/create my own rotation but I guess Im not virtuosic enough in programming/math/logic to design a proper healing rotation. The main problem I see with healing-simcrafts is that rotation are way to static. Raidhealing is about anticipation, awareness and good decisionmaking. Apart from the obvious boss-knowledge you get way better results knowing how your raidmembers like to play (e.g. who will fail more likely? Who uses defensive cooldowns or selfheal?).

Would be interesting to know how complex it is to add a new boss. For example healing Cenarius is way different from Ursoc, both are very good for druids but you have to prioritize different spells. A computer can't know that. Not asking you to make an A.I. but can you make an A.I.?

Cheers
RestoSpirit

[Yellowfive]

Keiyra I was just making a joke more than anything to emphasize the main point of the thread: you can flip your logic around and point out how using a strategy that prevents a resto druid from maximizing the gains of mastery is leaving a lot of potential healing on the table. Healers need to play together to maximize every healer's potential rather than playing the "meter-snarfing" mini-game.
@RestoSpirit The healing rotations are a bit more complicated to make than DPS rotations, for sure. Also, our goal is not to try and make a full "AI" or try to perfectly model every raid situation, that would be insanity. We see healing simulations as a way to explore theoretical possibilities and better understand what kinds of strategies could work well in different situations.

Even without making some perfect model of a fight or a player, you can get interesting information out of controlled healing tests in a simulation. It can give insight into how well a particular spell pattern works, or how stats interact with a specific healing style. We're going to make a lot more rotations and scripts to test different things, and we are of course always open to suggestions for things to try out, and you can customize it all as well.

A lot of people base much of their healing theorycraft on e.g. spreadsheet calculations to do the exact things mentioned above. It is boggling that someone would think a spreadsheet gives better data than a good simulation, even if the simulation is using a simplified model of a raid situation and how a healer plays. Even a simplified simulation is an order of magnitude more interesting and more likely to give useful data than a spreadsheet. The number of assumptions that a spreadsheet needs to make in comparison to a good simulator is much higher, and thus much less likely to give actionable data.

So for the spreadsheet healcrafter, think of this simulator as "the ultimate spreadsheet, on crack."

[RestoSpirit]

We're going to make a lot more rotations and scripts to test different things, and we are of course always open to suggestions for things to try out, and you can customize it all as well.

Im looking forward to it, for now the simulations give much lower HPS results than what I can actually achieve. Probably because of the fact that the simulation runs out of mana or is not using artifact-weapon synergy to it's full potential. Anyway simulations are great, I much rather have a few hours of simulations going then running 50+ heroic runs of the same boss to find the best healing strat for mythic.

Cheers
RestoSpirit

[nectarine]

Designing a model (and then basing stat weights) with the assumption that players will adapt to it, rather than just modeling how people actually play as a team, after a decade of refinement (i.e. the general rejection of strict assignment based healing in raids) in inherently flawed. Trying to reconstruct the dynamics of how healing teams work to that degree, in order to validate your results, just seems a little naive. It is no secret that mastery COULD be powerful if everyone played around you, and you played specifically to its strengths, but when it comes to actually progressing, that just does not work in a team, there naturally has to be overlap for everyone to be at their strongest potential simultaneously.

You should take some time and play with the RDSW Weakaura across different formats to see how they perform in a realistic environment where you are part of a team and not molding everything to your advantage.

The link for the RDSW weakaura isn't functional. Would you mind uploading it for me?

Thank you~