Resto Druids: Haste vs. Mastery
Posted on by Hamlet
With all the various things I focus on lately, one thing I haven’t been doing often enough is giving play advice and analysis of my favorite class, Resto Druids. Today I want to give a detailed discussion of one of the topics that has gotten a lot of attention lately–whether our most favored stat should be haste or mastery. In my Resto Guide I say haste, but don’t have the opportunity in that format to explain the recommendation in detail. Here, my goal is to work through the question very thoroughly, answer your questions, and see if I can help get you comfortable with a stat decision for this expansion.
Mastery increases the bonus from Harmony by 1% per 88 rating. It is additive with a flat bonus of 16.25% from the baseline effect and raid buff. Measured against a starting point of having no mastery from gear, each relative 1% increase requires 102.3 mastery rating. So long as the buff is maintained, it improves all healing other than the Lifebloom bloom, Ysera’s Gift, and Dream of Cenarius.
Haste has two effects. First, it increases the tick frequency and therefore the total healing per cast (I described the mechanics details here) of any HoT effect by 1% for every 90 points (as of this week). With Resto’s attunement taken into account, this is 1% every 85.7 points (haste rating is not additive with anything). Second, it reduces the time it takes to cast all spells by the same amount.
Breaking Down Haste
Because so much of Resto’s healing is in the form of HoTs, haste, above and beyond its cast time and GCD reduction, directly adds healing done to many of our spells. For most other healers this is a feature generally reserved to non-haste stats, which contributes to my low view of haste for other healers. For Druids, most heals go against the general rule and increase their output with haste like they do with other stats. This includes Rejuvenation (except for initial tick), Wild Growth, Lifebloom (except the bloom) and Wild Mushroom. It also includes a few rarer spells, Force of Nature and Dream of Cenarius, which are essentially throughput increases since they get cast time reductions with no related mana cost.
The basic analysis of haste in this post will be done by keeping the two components separate throughout, and remembering that haste is as strong as their sum:
- The cast-time reduction, which applies to all spells, but does not improve healing per cast. Therefore, it increases HPCT (healing per unit cast time), but not HPM (healing per mana), of all spells, by an amount equal to the haste percentage. Because some spells have cooldowns and some don’t, it only increases the cast frequency of certain spells.
- The value of added healing on certain spells, which increases HPCT and HPM to those spells and does nothing for the others.
The result is a complex mosaic of effects, unlike Mastery’s totally uniform HPCT and HPM increase to nearly all spells. Some spells have their healing per cast increased and not the rate at which you cast them (Wild Growth), so the effect is parallel and easily comparable to mastery. Some spells don’t have any healing per cast, HPM, or frequency increased at all (Nature’s Swiftness-HT). And on the flip side, some get an HPM increase as well as a double-dipped HPCT increase (Rejuvenation). So there’s no simple comparison; it will largely come down to the frequency with which you use the various types of spells.
So let’s try to break our typical spell usage into a reasonably small number of categories that will help clarify this.
1. Spells where haste increases the healing done with no effect on timing or usage (i.e. spells where haste and mastery do the same thing): Wild Growth, Wild Mushroom, and Lifebloom.
These are the easiest spells for which to evaluate the comparison. Haste and mastery are essentially interchangeable in their effect, and haste has a significant numerical advantage in how much it does (at 0 rating each, haste is 20% more effective, referencing the rating comparisons above).
One other spell that’s best put in this category is Clearcast Regrowth. It doesn’t sound like it fits, since the frequency of procs increases, but not the healing. Since it’s a free spell, the combined result is the same though. As you add haste, you do proportionately more healing with CCRG per minute, at no added mana cost (in fact, at no added time cost either, since the shorter cast time and higher frequency cancel out).
2. Spells where haste has no effect on healing done: Tranquility, Swiftmend, and NSHT.
These are also easy to evaluate. Haste does nothing of significance for these spells. Their use constitutes the argument against haste, as this is the only place where haste is worse than mastery.
3. Spells which are like the spells in category 1, but with the further benefit of being able to spam it more frequently when you choose: Rejuvenation.
This one gets its own category because it’s behavior is unique and because it’s a centerpiece of healing in all circumstances. A lot of the final outcome rests on the evaluation of Rejuvenation.
I started to give it away in the heading. Haste’s effect on Rejuv is at least what it is for the spells in group 1, because it has the same healing-increase effect, before you even get to the ability to cast more often (if you’re doing really meticulous math, technically the initial tick–around 1/7 of the healing depending on stats/talents, should be excluded from that).
The Effect of Faster Casting Speed
What to make of the increased cast rate? The easiest way to view is increased flexibility. You can choose to cram more Rejuv casts into, say, a 10-second period. The added casts are of course not free, but doing them in less time is a valuable option. Keep clear the distinction–haste is giving 2 separate benefits to Rejuv: the added healing per cast, which is free healing, and the added casts per unit time, which is not. One benefit is positive HPM and the other is neutral to HPM.
The total effect is best seen with two opposite scenarios. In the short, hectic fight, where healing is under significant time constraints and mana is secondary (Butcher), haste gives benefit to all the spells mentioned above and a double benefit to Rejuv, making it extremely dominant. In the long fight where efficiency is paramount (Mar’gok) cast speed is almost irrelevant, because taking a little longer to cast some rejuvs doesn’t matter–you have all that downtime anyway. There, haste is worth its basic benefit to all the spells mentioned in categories 1 and 3, as was discussed. The comparison to mastery will be based on what % of your healing is done with these spells.
Most fights have some of both elements mixed in. The comparison becomes clearer when you add some numbers, though:
- Where HPM is the sole concern, haste adds 1% per 85.7 rating to spells in categories 1 and 3. Mastery adds 1% per 102.3 rating to all spells. This means that if around 1/6 of your healing is from category 2, the two stats are equal. If category 2 grows, the value of haste will decline gradually, but the stats will be, loosely speaking, similar in value.
- Where speed is important, haste is much better than any other stat. In the extreme where you truly don’t care about mana, Rejuv scales doubly with haste, and instead of haste being 120% the value of mastery, it will be 240% the value of mastery on this very significant spell (when raid healing that rapidly, most GCDs are used on Rejuv). In a moment where bringing people up is a goal worth spending mana on, which, to put it simply, is not very unusual, haste blows away every other stat including mastery.
I’ve been working in numerical details gradually while focusing on explaining the thinking, but this was a key step. With no numerical context, it’s one thing to talk about “efficiency vs. throughput” in the abstract. And if you stopped there, you would very compelled to say that the importance of efficiency in Warlords would lead you to mastery. But what matter is that haste is slightly worse for efficiency (and in some cases, possibly not even worse!), and far better for increasing your throughput ceiling. It essentially free flexibility to do much more healing very rapidly when you need to, without significant hit to your efficiency even when you don’t need to.
We’ve seen that if category 2 spells are around 1/6 of your healing, haste and mastery are equally strong on an HPM basis. This requires us to revisit the first open question from above: whether to count Tranquility, a spell which can by itself account for 20% of your metered healing. On the one hand, healing is healing, and it often is that simple. Healing done by Tranquility, even beyond what’s needed to survive in the moment, saves on healing required in the subsequent seconds. There are 2 reasons you might discount its value:
- Healing added by Tranquility is much more likely to be overheal than most other spells.
- Concentrating healing in a 3-minute cooldown is worse than improving it uniformly.
The first is probably true, but I’m always hesitant about “it’s probably overhealing” arguments, probably because they can be applied to anything. This discounts the value of Tranquility, but maybe not by much. The second is more important though. Even if the Tranquility adds a lot of points of healing as it grows slightly larger, does that reduce expected deaths more than improving my healing during the other 174 out of every 180 seconds? I think not. There’s a certain diminishing returns on dumping that much healing into the raid at once. The meters, which count all points of healing as equal, don’t show it, but adding, say, 50,000 healing on top of the few million poured onto the raid during a Tranquility is not as likely to avoid a death as 50,000 healing done at some other time when the raid has not just been powerfully stabilized.
You can use your judgment, when looking your heal breakdown, of whether to exclude Tranquility, or include it, or count it for a partial amount such as 50% (which seems reasonable).
Putting it all Together
I avoided, in this post, simply building a model and throwing it at you (which I think is actually one of the least effective ways to present a theorycraft result). The goal was to build up the thinking with only the math required for important context, and see how far it can go (it turns out, quite far). I’m not hiding the ball on any math–if you want to dig deeper on your own numbers, go look at your logs from whatever fight you’re working on, and see what % of total healing is being done by category 2 spells, and keep that open while you read this section. That’s Tranquility (keeping in mind the previous section), Swiftmend, and HT (if you use a lot of non-Clearcast Regrowths you can include them too, but if you care about HPM, you should probably stop doing that in the first place).
I don’t have your log in front me, but I would be surprised if Swifmend and NSHT combined to count for more than 10% in any normal situation. Tranquility likely will push that over 20% or even 30%. That’s why it got a section to itself–in typical cases it will singlehandedly flip the comparison of haste and mastery for total HPM purposes. There’s a very real question of to what extent, if any, you actually want to “gear for Tranquility.”
Looking at your spell breakdown and making a judgment about Tranquility, you will have a good stat comparison on a pure HPM basis. This will either have haste ahead, in which case your decision is done, or mastery ahead, in which case you have to think about whether you ever care about anything other than HPM. Because, as described two sections up, that would make haste a clear winner if they’re otherwise close. I think it’s the rare fight where you should take the position that you don’t.
Math Interlude: Scaling
One topic that I don’t have nearly the space to treat properly in this post is how any stat gets relatively better as other stats increase. In other words, if stats A and B are equally-valud when you have 0 of each, and then you stack a large amount of A, B gets better relative to A. People bring this up a lot in stat discussions, but it’s much easier to trot it out than it is to weigh it carefully. And used injudiciously, it becomes a way to avoid reaching a firm conclusion on stats by always falling back on “balancing” them. I’d caution heavily against relying on this effect to push back on stat priority conclusions unless you really know what you are doing mathematically.
For a brief look–I said above that, at 0 haste and mastery from gear, haste’s numbers were around 20% stronger (to any spell affected in the same way by both), based on the rating conversions. This number will be very close to that any time you have equal haste and mastery rating. If haste starts to far exceed mastery, its value will decline, but not rapidly. For example, for haste to decay down to the match the value of mastery, at 0 mastery (a relative 1% gain per 102.3 rating), you’d need 19.35% haste from rating alone, or 1742 haste rating. And of course, you have far from 0 mastery on your gear, it being your second-best stat. It takes very, very wide stat disparities to change the sorts of results described here. If your haste is ~1000 points higher than your mastery, then yes, the HPM equalization will occur at around 8% healing from category 2 rather than 16%. That effect is real, and I wouldn’t mind keeping an eye on it in future tiers if gear choices happen to allow for overwhelming haste stacking. But once you account for all the factors discussed in this post, so far it is not likely to change the outcome in typical cases.
While the problem is an interesting one, and it was a great chance to touch on lots of points of healing theory in the process of a practical problem, the goal was to have a clear answer for people. And from my perspective we do: there is convincing support for haste as a generally recommended stat for Resto Druids.
With the analysis of this post in mind, we can construct a situation where you’d prefer mastery. Generally, one where:
- HPM is your only concern (you are never in a rush to top people off and always have ample time to use high-efficiency heals),
- Tranquility is of such importance that you’re inclined to alter your setup for that spell alone,
- You’re not using Dream of Cenarius, and
- You’re maintaining 100% Harmony uptime.
The latter two are standard assumptions that seem odd to call out, I know. But full-blown endurance tests, the situations we’re talking about, are precisely where they might fail to hold, and either of them can somewhat undermine the value of mastery.
I don’t deny the possibility of such a situation, but it’s an edge case at best. And where it applies, the edge to mastery is slight. The most serious players reading this can feel free to keep the possibility in mind, and hopefully this post arms you with a great deal of detail in making the evaluation of when it arises.
But the focus is post was identifying whether one stat was much better most of the time, and I think it’s done that. For the many people whose question was simply “haste or mastery?”, my answer for Resto Druids in Warlords is: “haste.”