Healing Theory, Part 2: A Tour of Your Character Sheet

My last post was a trip back to the fundamentals of healing, and claimed to be the beginning of a series where we build from that to more detailed analysis of the various healing classes. There’s still a little ways to go before we’re ready (both in terms of my math work on healing classes, and in terms of laying the foundation in posts) for very detailed cross-class comparison or balance discussion. But let’s get a little less abstract than the first time around, and look at some stats. Along the way we’ll not only relate back to the principles of the previous post, but finally have some foundation to approach the throughput and regen questions of the earlier Int vs. Spirit post.

Spellpower

All heals scale linearly with spellpower–there’s a base heal amount and then a spellpower term with some coefficient particular to each heal. An interesting point in MoP is that for nearly all heals, the base amount is scaled to be roughly 11,000 times the coefficient. For example, Divine Light has a mean base heal of 16817 and a coefficient of 149%, a ratio of 11287. For the HoT portion of Lifebloom, the base heal of each tick is 9315/15 = 621, and the coefficient is 5.7%, for a ratio of 10894. I actually don’t know of any heals off the top of my head that don’t follow this, but there are probably a few.

The significance is that heals tend to scale in proportion with each other as your spellpower increases. With 11000 spellpower, they all do twice as much as they do with 0, and so on. Since for raiding purposes, your spellpower is affected by a 10% buff, the better rule of thumb is that the base heal is worth 10,000 spellpower. This is handy to keep in mind, as it gives you a concrete picture of what a spellpower increase means to you. If you had 20,000 spellpower when you started raiding and now you have 30,000, you can expect that all of your heals are healing for 1/3 more than what they were before (and that’s before mastery scaling and any other benefits you might have).

An interesting footnote is that for DPS spells, the number seems to be closer to 2000 or even less, rather than 10,000. In general they seem to want the steady increase to healing spells over the course of an expansion to be much less dramatic than DPS.

Intellect

The main function of Int is to provide spellpower. The only additional note that applies to Int specifically is that it scales 5% better than spellpower due to your armor specialization or equivalent cloth bonus, 5% better again due to raid buffs, and potentially more if you have talents like Heart of the Wild.

Int also provides chance to crit, although it’s not very much. Consider the crit contribution of 1 Int to be about 1/4 of a crit rating. Spellpower, being a primary stat, should generally be worth about twice as much as crit rating, a secondary stat. So Int’s spellpower contribution is an order of magnitude more significant than its crit contribution.

Mastery

Now we start getting into something interesting. 600 mastery rating gives what was called, before MoP, “1 Mastery.” They don’t show that on the character sheet anymore, but it’s still convenient to compare mastery bonuses in terms of the % increase for each 600 rating.

Monk: 1.2%*
Druid, Holy Priest: 1.25%
Paladin: 1.5%
Disc Priest: 2.5%
Shaman: 3%
*Monk is irregular since the mastery bonus is then multiplied by an individualized modifier for each spell, and for other reasons.

Druids have the simplest healing mastery. Harmony is a flat % bonus to all healing, so long as you have cast a direct heal in the past 20 seconds (in practice, this has 100% uptime). So that’s an interesting starting point. For the most boring possible bonus, the uniform % to all throughput, the standard of mastery scaling is 1.25%. Not much more to say there.

Holy Priests are similar, getting 1.25% via Echo of Light, applying to all heals besides Renew ticks, Holy Word: Sanctuary, and Lightwell. Maybe we’d have expected a small bump upwards to the bonus since it excludes a few heals, but I guess they thought it was a small enough portion that it won’t throw off scaling. We might also expect the lingering HoT to be a slightly less effective heal delivery than the Druids’ straight-up increase, or the shield-oriented bonus of Paladins and Disc Priests.

Disc Priests are interesting: fully double the usual scaling, applying only to PW:S and Divine Aegis (looking at the 5.2 version that excludes Spirit Shell). Without conducting a study of Disc logs since this post is a survey: if PW:S and 50% of the value of crit heals and Prayer of Healing collectively account for about half of what Disc Priests output, then their double-strength mastery scaling will be roughly in line. Disc is also a good case study for how this is sort of high-level scaling analysis can predict balance problems. 5.1 Spirit Shell by itself gave full mastery scaling with up to 25% uptime, controlled by the player, stackable with other buffs, and in shield form. This is definitely suspect for a mastery bonus that shouldn’t be affecting more than about 50% of total healing even if it were randomly distributed.

This is the sort of analysis we might hope to do in more detail when getting into meaningful questions of comparison across all classes. Shaman get an interesting quick note, since their 3% Deep Healing balances against a blanket 1.25% mastery if typical heal targets are at around 58% HP. Monk mastery is the only one that’s not a loose variation of a flat % to some subset of your heal spells, so there’s no general way to predict how it will scale without going down the whole spell list. I imagine they had to balance it mostly ad hoc, and similarly, evaluating the size of the bonus for Monks can’t really be done without quite a lot of arithmetic.

Crit

Crit increases most heals for most classes by 106% with a meta gem. Its scaling is generally weaker than mastery 600 at rating for 1%. Discipline effectively has a (300+Mastery)% bonus on non-PoH heals due to Aegis and in fact crit outscales mastery quite heavily for these heals. This effect might be more noticeable after the 5.2 Spirit Shell changes.

There seems to be a widespread mental slant against the value of crit for healers because of the chance of overheal (or a slightly related argument that I think has a little more merit, the inconsistency of it). I’d be careful about writing off the value of crit in this way. I don’t actually know of any good evidence that increased critical healing cancels out most of its effect by increasing your overheal ratio (this would be a good log study). In general you want to watch out for the “it’s all overhealing” argument, because it can be applied to pretty much any healing stat (and tends to be in most forum discussion) without much reasoning beyond that fact that it’s easy to picture in your head.

Most heals overheal quite a bit–look at any log. The question when adding to your healing output in any way is whether the new healing is likely to, on average, have the same efficacy as the prior healing, or more, or less. Maybe you’re inclined to argue that crit heals must have lower effectiveness because they can only be effective if the underlying heal would have been fully effective. But this argument generally applies to increases from spellpower, crit, most masteries, HoT breakpoints, and (slightly modified) haste. It’s neither here nor there.

Short version: added healing is what it is. Even though increasing your total raw healing output isn’t itself a goal the way it is for DPS, one can generally expect that small increases in raw healing, other things being equal, will manifest as similarly-sized increases in effective healing.

Haste

This is where we get a little less mathy and have to think back to the principles again. The math of haste is not mysterious at all: 1% to everything you do for every 425 rating. But let’s pause at “1% to everything you do”:
–cast time of anything is reduced by 1%.
–you burn mana 1% faster while taking advantage of this.
–your cooldowns aren’t changed.
–HoT spells might get extra ticks, but to avoid any quadratic scaling, they either have a constant cast time (Rejuvenation, Renew), or have cooldowns, so they get no other benefit (we should make note to further examine Riptide in this regard).

So the first red flag is that for spells with cooldowns, haste is positively unexciting. This is important because the relationship between cooldown and non-cooldown spells is an important one that’s going to be developed more in later posts. Generally, healing class design tends to have some strong, efficient spells delineated by coodowns to establish a sort of baseline amount of output, and then spells you can use in the remaining time that can be modulated more widely based on healing need, time, and mana. The stratification between cooldown and no-cooldown heals isn’t totally strict (in particular, Prayer of Healing really pushes some boundaries). It’s an interesting pattern though because, like the Int scaling, it appears designed to reflect a desire to dampen the scaling of healing classes as gear increases.

End digression. Haste doesn’t let you cast Circle of Healing or Holy Shock or Spirit Shell more often. In fact, whatever effect there is on cooldown spells, haste generally lets you use them less often, assuming you actually make use of the haste to spend mana on no-cooldown heals more often.

Haste is hard to make a case for in specs that, like healers, have dead time in their rotation. People tend to point out the value of haste in burst output. My basic problem here though, is that nearly all stats (other than regen) increase your burst output. Mastery (where it applies), crit and spellpower all give you that added burst HPS at 0 mana cost, while haste gives it you at mana cost increases in full proportion to the added spells you cast.

So we’ll see where we wind up in terms of haste after a closer look at class mechanics, but it’s an uphill battle.

The one completely different effect of haste is HoT breakpoints, which do function like other stats in that they increase healing per cast. You have to be careful not to overempasize this though. In the long run, haste increases healing per HoT cast by an amount equal to the % haste increase, nothing more. So the clumpiness of the haste benefit to HoTs can be used to your avantange when you are close to a breakpoint and can get a chunk of healing for a few extra stat points. But once you’re there, you have to pay full price to get the benefit of the next breakpoint anyway. So you should know your breakpoints to make the proper small-scale adjustments, but Druids are the only healer who use HoTs so much that haste is a “true” increase to a significant portion of healing.

Spirit

This section is a bit of an outline or placeholder; a full followup to the Int/Spirit article will be a post of its own. Here are the most salient things to keep in mind.

Spirit does not let you cast your efficient cooldown-bearing heals much more. It might, to the extent you reduce use of them somewhat when really strapped for mana. Mostly though, this means you weren’t budgeting mana well when you had less of it, if you weren’t already using your most efficient heals as much as you could.

So in some ways it’s similar to haste in that Circle of Healing-type spells don’t get much out of it. The benefit of added mana accrues primarily to whatever you freely cast in between, which you can now do more. Even without making assumptions on how you cast, at least take in the concept: the value of Spirit depends on which spell you will actually cast more with each marginal point. If that spell is an inefficient filler like Regrowth, you might be better off doing what you can to bolster other parts of your healing rather than relying on it more. If your mana->throughput factory is Prayer of Healing, maybe we’ll find that you can take that Spirit all the way to the bank.

One unique Spirit mechanic in MoP is Rapture. At first glance it’s worth 200% Spirit every 12 seconds at most, or 0.83 MP5 per Spirit. This is perhaps a bit concerning since the entire value of Spirit otherwise is 0.56 MP5 per. The reality is trickier than that though, since it requires an investment is PW:Shield, which is not particularly efficient. But given that Discipline Priests are already trained to make use of the mostly-free shield, the net mana on their balance sheet really does increase by 0.83 MP5 per Spirit added. Even though this post is only a preface to full-blown class balance analysis, we have definitely found something to put in our notebook for later. For some reason when Spirit was standardized for all classes, Disc Priest was left with a very strong Spirit scaling mechanic that more than doubles its value, and intentionally or not, they’re the same class that can pour every point of it into the PoH machine.

Total Mana

Most classes have a mechanic that returns mana based on a % of your total mana pool.

Druid: 20%/3min from Innervate.
Paladin: 12%/2min from Divine Plea
Priest: 27%/3min from Shadowfiend or ~16%/min from Mindbender. After the patch, another 6%/min from Power Word:Solace may become much more popular.
Shaman: Nothing outside of the net gain from Telluric Currents, but they do get at least 2138 MP5 (fixed) from Water Shield, and a discount on crits with Resurgence.
Monk: 1% for each Chi spent from Mana Tea. Can be worth upwards of 20%/min.

Nothing too deep here, but I’m kind of surprised this isn’t a little more uniform. I’d imagine that that class balance would be a little easier if the high-level mana economy was generally expected to be equal for everyone (but still provided by different class mechanics). It seems like unnecessary work to give Monks hundreds of thousands of mana with Mana Tea over the course of a fight and then have to balance the class around the increased resource pool.

The only material thing to add here is that Ember meta is worth 6000 mana times the benefit listed above for your class. So Druids and Paladins are looking at around 30 MP5, Priests 80 MP5, and Monks around there or a bit higher (minor effects like Hymn of Hope don’t change this much). All of these amounts are a bit lackluster next to even a modest crit bonus; the mana-oriented meta needs an update to reflect MoP mechanics.

One thing useful to keep in mind is in total, where you get the mana that you spend over the course of a fight:
300000 starting. In a 5 minute fight, this is equivalent to about 5000 MP5.
6000 MP5 constant base regen.
Above amount, up to around 5000 MP5 depending on class from class-specific mechanics.
6000 MP5 from Spirit if you have about 11000.
Again this was very rough since I didn’t want to break it down by class in service of a small point. The point is that most of the mana you get to spend over the course of a fight is quite fixed, with the Spirit slice really only occupying 1/4 or 1/3 of it. The bulk of your mana is from the constant sources you tend not think about and aren’t much affected by gear.

Conclusion

This was the framework for looking at healing spells and abilities across all classes in more depth in the future. The high-level overview of each stat and how it behaves is context for what we might expect out of different spells and how they should scale with gear. The next step will be to start digging into classes individually, and seeing how they fit into the expectations and patterns described above, when taking into account their various mechanics in full detail. Only then will we equipped to look upon the entire picture of WoW healing at once with sufficient perspective to tackle complex problems such as class balance.

11 thoughts on “Healing Theory, Part 2: A Tour of Your Character Sheet

  1. Great intro. I look forward to reading the in-depth analysis.

    I think you mentioned it, but I’m hoping that it’s possible to model absorb healing and reactive healing (e.g. Living Seed) somewhere in this analysis. I’m curious how a healing priority order (e.g. absorbs first) is balanced among inhomogeneous classes.

    • Depends on what you mean my balance. Think back to the arguments about meters back in the first post. In a whack-a-mole contest, shields always win. So shield-heavy classes are disproportionately successful at meters.

      Which is not to say that their real benefit to the group is low, just that meters exaggerate it. The benefit of shields is actually quite high, since they’re mostly heals that come with some free HP buffering.

      Ideally healing classes would all have some damage reduction component, because the inhomogenity there is difficult to balance. There’s some attempt to do this with Druids having Living Seed, the Paladin mastery, and Shamans’ Ancestral Vigor, but these things are all of wildly different magnitudes.

      So I also don’t entirely like the mentality that most spell idea that involves damage reduction are focused on the “shield specialized” healer. They get some of the best cooldowns by default (Barrier, Pain Suppression), because the DR/shield fits their flavor. I’d rather all healers have protective tools to a more uniform degree.

      • At what throughput level does the argument “shields win in a whack-a-mole contest” turn into “shields are unbalanced”?

        Across all 25H logs (all parses, not the top 100), raidbots reports that the median disc priest throughput is 30% higher than resto druids (62,196 vs 47,906).

        Don’t get me wrong — I think that a systematic analysis of healing classes is really a fantastic idea. But healing is a team effort. If one member brings 30% less throughput, what additional benefit can they bring to make up for that deficit?

  2. One small aspect of haste you didn’t touch on is its value in reducing reaction time for healers. In a more static situation you’re right it has virtually no value, but in a whack-a-mole dynamic healing situation shaving tenths of a second off reaction time can translate into live versus dead raiders.

    Admittedly there’s probably more room for player improvement than the available gain from haste, but it is still there.

    • I’ve always been a bit skeptical of this. For one, “tenths of a second” is quite a lot of haste. For something like a typical 2 second heal, taking off two tenths of a second is around 5000 haste. For smaller changes in haste rating, the effect is much smaller.

      There’s also some best-case scenario mentality here. Haste can save someone they would have happened to take a lethal blow in the tiny fraction of a second by which your heal is shortened. It’s actually quite a coincidence for that to happen. Most of the time the damage will come at any other moment and the stronger heal will give more benefit. Also, on a full-HP target the situation in reversed–a heal that comes right before a hit will overheal and one right after the hit will be effective.

      I just wouldn’t get too caught up constructing mental scenarios that fit too perfectly. Haste increases cast time by a certain percentage, and only roughly that percentage of hits will happen to be right in the window affected by haste, rather than some other time during the cast (unless you think there’s some strong nonrandom factor makes damage very likely to happen right around the moment a heal lands).

  3. I’m really enjoying this series. I enjoy more rigorous looks at healing, and starting with a framework from which to work makes a lot of sense.

    I’m not sure if you plan to further address the common healer complaint against critical strike (“Healers should avoid anything that’s ‘RNG’ “) but it seems to me that, whether or not inconsistent returns is bad, it still depends greatly on the spells in question. For example, a holy priest casting Circle of Healing, Prayer of Healing, Renew and Cascade will be looking at tons of “spell hits” per cast, which means that critical strike doesn’t require the long sample of an entire encounter to “even out”…it should even out almost on a cast-by-cast basis because every spell cast consists of a relatively large sample (rather than a single did-it-crit-or-didn’t-it boolean like a single-target heal).

  4. Pingback: Disc Priest - Mists of Pandaria - Page 24 - Elitist Jerks

  5. Pingback: Healing Theory, Part 5: Haste | It's Dangerous to Go Alone

  6. Pingback: Healing Theory, Part 6: The Mana Economy | It's Dangerous to Go Alone

  7. Pingback: Warlords of Draenor Mini-Analyses: Healing Gameplay | healiocentric

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *