Aside from you previous post do you have any factual proof that shows the value of a crit heal being 1.36 (136%) or is this based only upon the things you mentioned (overheal/mistiming) in your previous post?

Nidaba, how are you enjoying having two points in Holy Reach and two points in Healing Prayers?
I am considering taking my Test of Faith and Desperate Prayer talents out and doing just that.
Cheers-thanks for all the wonderful information.

A very good point as well. For discipline priests multiple factors converge to improve your HPM when you increase the size of your heal since in addition increasing the numerator (heal size), the denominator (mana cost) is indirectly decreased by mana returns from rapture.

The big factor is probably how many other healers you have, so there would be a difference between 10 and 25 mans and the gap between overhealing in non-crits and crits would probably narrow as you add more healers (at the same time overhealing goes up). I do not recall the exact numbers I had for overhealing but both non-crit and critical heals were overhealing 40-50%, which indicates a significant degree of interhealer interference/competition. Probably that kind of environment is exactly where you'd see more heals either completely landing or completely overhealing.

Something I have not done (and maybe should try) is measuring overhealing in non-crit vs crit heals in 5 mans where I am the only healer. I would guess that there I might see a more significant difference since (hopefully) non-crit overhealing should be quite low.

By the way, I like partnering with a druid in 10 mans in part because druids and discipline priests don't interfere with each other very much. I'm sure I still snipe some hots with flash heals and penance, but shields play well with hots.

Edit 1: I've looked at more data on overhealing in crits versus non-crits, and I'm seeing a lot of variance. I've seen differences below 1% and as high as 16%. This is all in 25 man raids, so I can't blame this variance entirely on raid size. In any event, that much variance makes it hard to make accurate predictions.

The 2 points in Healing Prayers are key if you use PoM on cooldown. Works out to one of the highest regen talents available to us; the fact that it reduces mana cost on PoH (which I use a lot) is just a bonus.

As far as Holy Reach, it's nice for CoH. I've debated getting Test of Faith, but we run with resto shamans, so we rarely have people below 50% anyway. It's a nice talent, but I'm not sure where I could justify stealing points from.

You'll note I have 3/3 Imp Renew. With that, max spellpower (almost 3k raid-buffed), and the Glyph ... I'm almost healing for as much as a Rejuv from a non-optimized druid. /sigh

I've had 3/3 Test of Faith for a while and while I'm never really conscious of the times the extra crit is being added in, I still like the idea that it's there (I wish the bonus was 12% instead of 6%, though). I haven't run into any situation where I need the extra mana from max Serendipity or Healing Prayers, and I use PoM nearly every time it's ready. Even on a fight like Sapphiron where PoM is 44% of my healing and I do 6000+ HPS, mana isn't an issue.

As soon as the mana regen outlook changes, I'll probably look to those talents more. Considering the changes coming in 3.1, I can only speculate that it might be possible to get more of the Holy talents we like without struggling with the 13 required for Meditation. I'll probably switch 2/2 Holy Reach for 2/2 Healing Prayers before I drop any from 3/3 Test of Faith, though.

Also, as I predicted, I did another 4-healer Naxx 25 this week and Renew was more useful than it had been in the past (because there is less sniping going on). It was up to 9% of my total healing, and was even higher on Malygos since I am back to using it during vortex. It ticks for about 2200 glyphed and talented, which is 8800 total healing for 656 mana. Its efficiency is okay IF it has a chance to heal, and it does when there are fewer healers in the raid or more damage and other healers have to focus on other things and can't afford to snipe other targets. I'm hoping Ulduar presents these sorts of opportunities.

As an experiment, I'm thinking it'd be fun to see how Renew would behave if it could a) crit, b) accelerate with haste, c) proc all other crit-based talents. It might be too much, but it would at least scale with more than just spellpower.

Stem those tears! who knows really where exactly this additional talent or two would end up though.

It took me a while to decipher my own calculations, but here is how I arrived at that figure:

I used WWS logs for a few consecutive raids from both myself and our other healers (a shaman). I examined each spell individually because I felt that how effect crits were would vary depending on the spell.

For each cast, I determined how much it would have overhealed were it not a crit. That is, if it was not a crit I used the actual amount it overhealed, and if it was a crit I used the amount it would have overhealed had it only healed for 2/3 of what it did in fact heal for.

I similarly determined how much each cast would have overhealed had it been a crit. Again, easy if it was a crit. Also easy if it was not a crit but overhealed for something, since I could just add 50% of the raw healing to the overheal amount. Much more difficult if it did not crit and did no overhealing, since it's not possible to tell for certain how much it would have overhealed for. I approximated this by calculating the percentage of my crits and non-crits that had *some* overhealing for the cast. By comparing these I could estimate how many of my non-crits would have overhealed for *something* if they had critted. Then, given no reason to suspect they would have overhealed for one amount or another, I used half the amount of the crit increase (25% of the raw healing).

When I had the figures for what the overhealing would have been with 100% crit and what it would have been with 0% crit, that gave me how much more healing would have been done had I had 100% compared to 0% crit. This is the same as how much more healing any particular spell does when it crits.

The reason I used this method rather than simply calculating my overhealing for criticals and non-criticals was probably partly because I'm insane. I also calculated the actual overhealing for both and the results that gave vs. using my complex method were not very different. I did notice, however, that by using my "what-if" typing thinking, the numbers seems more consistent between different spells. It's like my data set was small so I abstracted a larger data set out of it, particularly in the case of the non-crits where it is possible to determine *exactly* what my overhealing would have been had I had 0% crit without any guesswork.

As I said, this was generated using logs from my own raids. I'm not sure how different this would be for different people. I was actually quite surprised as how similar the number was between different spells, particularly that it was similar for PoM and direct heals. The only spell that really stood out as different was PoH, which makes a lot of sense because I use PoH primarily as a top-up type heal.

I hadn't really thought about how playing with a larger or smaller number of healers might have a significant impact on this number, but from what Promethia said about only noticing a 1 to 2 percent difference between her overhealing for crits and non-crits, it seems like it makes quite a substantial difference (which makes a lot of sense, since neither discipline priests nor shamans casting direct heals (as opposed to chain heals) are ideal for top-up type healing, discipline priests would probably shy away from it if there were other healers to cover that role and there were hots to avoid stomping). I'd like to do this test in some five mans as well, since, thinking logically, it does seem like crits should be wasted on overheals even more in that setting.

The healing it would have done if it had crit is (at least for Shaman) also depends on the strategy used to determine when to heal. I actually heal assuming that my spells will crit (43% rate), which would be a mess with gheal/HL, but when you're tossing a .8s 5k hit/8k crit heal then not critting someone 6-7k down only leaves them 1-2k short, which the CoH/WG or AA splashes cover. It's the same with a 1.5s HW 8k hit/11k crit for people 10k down. People at 4-5k down get a Riptide. I don't know if your Shaman does this, but I've been doing it since the expansion changes whenever everybody doesn't absolutely have to be topped off.

As for overheal with number of healers, it has a huge impact on my healing. With one other healer who's tendencies I know, I'm consistently getting 12-15% overheal range for the night. By comparison, I'm 20-25% overheal in a guild 5 healer 25 man and up to 30% for a pug OS/Archavon. The difference is almost always in healing collisions.

I do know for sure that overhealing is very strongly affected in going from a 5 man to a 25 man. My own overhealing at least doubles and sometimes triples. Because I'm a bit of a control freak, I tend to like the 5 man healing environment more in many ways because I have a lot more control over everyone's health. They may pop a health potion or something like that and mess me up, but otherwise I have a great deal of control over prioritizing heals in a way that keeps everyone up in an efficient way.

Part of that process is choosing the right-sized heal for the job. Frequently that allows me to know a flash heal isn't going to overheal at all since I know it won't take the target to full health. However, I'm generally not expecting a crit heal, and I don't know how much I hedge my heal size to account for possible crits. So I'm sure my crit heals would overheal more frequently than non-crits, and the only question is "how much?"

In a 25 mans, health bars are often moving very frequently and in both directions. It becomes very difficult to predict what the right-sized heal will be when the heal completes. It's often zero. Crit certainly introduces even more variability, but everything is so random that crits are not your major problem. I worrying less about crits and instead focus on guessing what other healers are likely to do. For instance, if someone (usually a dps) unexpectedly takes a "conspicuous" unexpected hit, I tend to ignore that because I know every healer in the raid just saw that and some of will surely be healing that target. Such a target will either die before anyone can do that or will get massively overhealed, so I often ignore that and try to heal someone else.

Anyway, I don't want to digress too much into how healing tactics differ in 5 vs 25 mans, but the point is that crit is a relatively small source of variability in 25 mans compared to interhealer competition. In a 5 man, there is no interhealer competition, so it seems reasonable to predict that crit would have a bigger effect on overhealing in 5 mans than in 25 mans.

In 25 mans i rarely if ever use my GH. It's all FH BH and CoH. Even if im stuck to a MT healing spot on a boss fight ill still use FH and BH and bairly a GH reason being because 9 times out of 10 you will be grouped with a palidian for the other MT healer. GH has no chance of standing up to their fast big heals. I heal off of crits, i just spam FH and when i know i need a crit ill pop a BH. I dont know the numbers but crits for BH seem to be much higher than any of my other spells ill take a 10k crit BH over a GH any day.

Factoring in the unpredictability of crit heals-- and therefore its propensity to be wasted as an overheal-- is an interesting question. (One with enough soft variables that I don't think there will be an algorithmic representation for it, but maybe we could find a way to get a better feel for it.)

If we're going to calculate the marginal throughput loss from a overhealing crit that wouldn't have overhealed without critting, we should also account for the marginal gain in efficiency from an additional Serendipity proc. Our most efficient heal isn't one that's completely effective; it's one that overheals by one. Therefore, a heal that would have been fully effective as a normal heal but overheals as a crit (the only case where crit increases overheal) loses effective throughput, but also gains the efficiency from Serendipity.

I'm going to take a first swing at constructing a model for that-- stop me if I blow any major assumptions.

For instance, let's just take l337n00b's 1.36 valuation as gospel for the moment. If we abstract it out, this means that (1.36/1.5) = 90.6% of critical heals are effective heals, and therefore that, on average, 9.3% of each critical heal is 'wasted' *from being a crit*.

The somewhat more difficult question is: on what fraction of critical heals does this wasted healing occur? Is overhealing a normally distributed function, where you usually overheal for about 10% of the crit, with overheals further from the mean occurring less frequently? Or is it better to model it as a uniformly distributed function, where you are equally likely to overheal for any amount from 1 point to 33% of the heals full value? My intuition leans toward the uniform on this one, but it's also usually a bad idea to bet against a normal distribution, so feel free to dispute me on that.

(Remember, we're only interested in the overhealing of the set of critical heals that would not have overhealed if they had not crit. Therefore the maximum possible overhealing for a heal of that sort is 33%-- the extra 50% of the crit's 150%. If it overheals for more than that, it would have overhealed without critting and therefore doesn't generate additional Serendipity procs.)

If an average of 9.3% of each crit is crit-based overheal, and the overheals are from a set drawn uniformly randomly from 0-33%, then the average overheal in the set would be for 16.5% of the heal's value. So crit-based overhealing would have to occur on an average of (9.3/16.5) = 56.4% of crits.

So, then, 56.4% of crits generate a Serendipity proc that wouldn't have otherwise existed. Therefore, the mana cost of these heals is (.75)(56.4) + (1)(43.6) = 85.9% of what it would have been without the crits.

If crits do an average of 1.36 times the heals of a normal heal, then they should do it for 85.9% of the mana-- which comes to 158% of the efficiency.

Which is pretty darn close to the 150% efficiency they would have if they didn't overheal.

So crit-based overheal makes up for itself, efficiencywise, with additional Serendipity procs.

Two points:
1) While I used the 1.36 value to whip up this math, I'm not sure it would be that sensitive to the starting ratio. More crit-based overheals means more crit-based Serendipity.

2) Also, I back-calculated the crucial percentage using some fuzzy assumptions, but it would be pretty easy to actually figure it out from the original logs. Just take the crits that don't overheal and the crits that overheal for less than 33% of the total, and get the ratio of overhealing crits to the number of all the heals in the set.

Right-- well, more precisely, you *could* make that calculation if you knew the underlying distribution of percent overheal. I made a guess--overhealing is roughly uniform random-- and worked from there. I agree that the distribution becomes much different on highly predictable fights with few healers like Patch 10 (probably something closer to a right-skewed bell) and, well, Patch 25 tank healing is the most extreme all/nothing case I can think of, but then you'd use a binomial random distribution.

I think the uniform random assumption is defensible in highly random situations like 25-member raid healing with multiple healers, particularly for the set of critical heals that overheal for between 0 and 33% of their value. It might also apply to more predictable situations, because, as a couple people have pointed out, the dominant factor in whether overheal happens is an all-or-nothing question of timing. For the set of heals we're interested in, though, that factor cancels out because the overhealing happens whether you crit or not. What I'm trying to estimate is the distribution of overhealing, not on all heals, but on crits that overhealed because they are crits. That seems both less predictable (and therefore not trending towards a normal distribution) and less all-or-nothing (and thus trending towards a binomial distribution).

But, like I said, it's easy enough to actually derive that figure empirically if you have the logs and the patience.

Blizzard has acknowledged the weakness of the "once beloved" renew, and I can't wait to see the spell buffed.
In my opinion, it is an essential element of the priest's toolbox.
We need a decent HoT if we are to follow the jack-of-all-trades concept. (This is an assumption of Blizzard's point of view on the class as a healer.) And besides, I want a hot, else I'd a paladin, wth!

On the other side, I get more concerned about GH now...
There's always more balancing to do.

I've seen more and more people let go of Test of Faith.
My thoughts on the talent are that it's a great tool when you hit hard, new content. But it loses it's effectiveness as the encounters get overcome.
I can still remember multiple 20k GH crits with zero overheal on our first 25-men patchwerk kill.


Since the topic has come to critical heals, I was asking myself how other priests feel about Surge of Light ?

I loved it at first! Especially since the start of the expansion featured expensive heals.
Although, now that we already have "decent" regen (nerf for Ulduar ?), and take advantage of our critical heals for throughput, I find the talent less and less appealing:

+ It is great in mana efficiency. This is a given.
+ It adds up with IHC to start regen nap. Redundant here.
- But SoL doesn't scale linearly with crit : procs will start to refresh themselves before use. Meanwhile, the HpM value of a normal FH also scales with crit via serendipity or simply effective healing. (Need mathematical comparison here.)
+ On the good side, the fights were you have the most SoL procs (saph, malygos, ...when there is a lot of raid healing basicaly) are also the fights were you use the most mana. Logical. (With the exception of sart+D, see further.)

+ It can be used on the move. Especially now that we have lost half our healing potential while moving with the CoH cd.
- But it's random. Fish for a SoL proc with a CoH on the raid while you move, if unlucky... renew... (Assumed CoH and prom are on cd and you're not regening).
- It's uneasy to save your SoL proc for when you know you will need it, because FH is now a main heal in our arsenal.
I think this may not have been intended by Blizzard when they designated the current SoL. But we don't cast many GH anymore.

+ The instant heal can save lives. Sometimes.
- PW:S too, if no SoL.
+ But PW:S is less efficient and can be a waste.
- It is hard to save SoL procs for emergency.
And it's obviously useless to snipe other healers.

- An IHC proc is far more attractive to me than a SoL proc.
Every time I get a SoL proc, I feel like I just lost a chance to get a critical FH, and a chance to get an IHC proc. This is especially true when you need to heal with reactivity a number of people in a short amount of time. Sometimes I wish I could toggle SoL off during damage bursts for increased throughput.
The best example of this is during the peak of Sart+D.

In short, I feel like the shortcoming of the talent are catching up to the benefits as I progress in skill and gear. And I wish it was something straightforwardly good. Like if you had a way to control the use of your SoL procs, or not give up throughput for it.

In a way, the old CoH was allowing us to bypass most disadvantages. (In an overpowered way, no denial. But seriously, CoH's glyph was a bad choice.) Or you could say the cooldown implementation transfered some throughput pressure on FH, when GH falls short.


Personally, I overgear a bit toward regen, because I watch more the health bars than my own mana bar (I like to top people off :s). Thus, I may be biased.
I will try a new build without it. Maybe force myself to use PoH more often.
I am wondering if some people also switched opinion about SoL after a point.

Oh, and just as a note : I never spammed CoH before 3.0.8
In fact, now that I assume it is not overpowered, and use it without any shame, I find it contributes to a higher percentage of my overall healing. >.<
Sadly it's not the kind of healing that saves lives anymore.

For the time being, I am not to excited about it. Ghostcrawler only mentioned 1 or 2 talents. And without a coefficient tweak or some stupidly overpowered talent, I can't see that being enough.

But it's a start. GC has also acknowledged that Priests might have become gcd limited in between cooldowns. That's also a good start. (Think I posted the quote in the other thread, not going to post it here).

So this might have been asked and answered before but i cant find it. How worth would getting the glyph of renew? As of right now i have it and my renew ticks for about 2k every tick. Iv heard that renew hasnt really scaled with other skills so getting it to heal more for less time is worth it right?

I switched Glyph of Dispel to Glyph of Renew in 3.0.8. I'm liking the scaling to my Renew ticks, and am trying to train myself to use Renew effectively as a supplement to my AE healing. If GC actually comes through for us and buffs Renew, I'll be in a good spot, habit-wise.

If you use Renew a lot (read: 10% of your healing over the course of a night), then by all means Glyph it. I've seen ticks average about 2150 since I got the Glyph, with my max throughput gear on. If I use a Frost Wyrm flask, ele shaman buffs, etc, it can break 2.2k.

The value of crit diminishes faster than normal heals with overheal. The main reason is that if a spell overheals even a little crit has 0 benefit. The thing to remeber is that x% overheal does not mean that (100- x%) of your spells have 0 overheal at all. Its a distribution and because healers try to minimise excessive overheal, its actually skewed towards the low numbers meaning even lower % of 0 overheal and hence an even lower value for crit.

1.3 is definately the corrent value to use assuming average overheal is between 15 and 25%. This is not terribly accurate, but using 1.5 is completely inaccurate, which is worse.

The value of crit for serendipity is also not linear and its actually very difficult to calculate. A rough sketch however shows that its much much smaller than your crit%, so 1% crit gives you a lot lower than 1% more serendipity procs.

Crit affects a lot of things, but not that much and it has pretty steep diminishing returns too. Its not a wasted stat, but its nothing special either.

===============================

I think renew is unjustly maligned. Unless someone is disc, its really not a bad spell. The renew glyph gives it a direct 25% boost to HPS and its high scaling with spellpower, means that you can now pull some very impressive HPS by just rolling renews. In terms of efficiency renew gets increasingly better than flash heal as spellpower increases. At 2k odd spellpower, the difference means that renew ain't really worth using over flash, but at 3k spellpower, it does make sense to use renew if it will have a chance to heal.

The main benefit of renew is using it to gain free time, in cases where someone takes regular damage. If someone is taking 4k ticks every 3 seconds and a pom at a later point, whereas with flash heal you really have to heal every tick to max serendipity returns. If someone is taking regular damage renew can be a good option, especially if you need to get ooFSR time.

The limitations of renew are of course well known and don't need to be mentioned further. I think if anything kills the value of renew its the CD on CoH.

Let's take a fight like Sapphiron where there's almost always something to heal. I did around 6300 HPS (raw) using PoM (42%), CoH (27%), Flash (22%), and Binding (8%). Now, my talented and glyphed renew ticks for 2150-2200 every three seconds. In the interest of round numbers, that's 700 HPS per renew. With the glyph, it lasts for 12 seconds, which means I could get it rolling on 8-10 people at once depending on haste/lag, but let's say I'm still using a GCD here and there for PoM and CoH. So even if I managed to use every GCD (100% inside FSR) for this and roll 10 renews at once (7000 HPS), I feel like I'd be limiting myself quite a bit. I think if you start talking about using not just renew, but -many- renews, you have to defer to PoM as a superior spell. That's not even including the goodies you get by ignoring renew completely in favor of spells that proc Surge of Light and Serendipity. Rolling renews just still doesn't compete for me, even when considering just raw HPS.

Start talking to me about Renew scaling off crit and haste and other talents, then we can really revisit these Renew-centric healing strategies and have some fun.


I'm pretty sure this forum was where I read that Renew became -less- efficient than a glyphed Flash Heal at around 2100 spell power. That implies to me that Flash Heal's efficiency graph has a steeper slope than Renew's. Can anyone with more math insight comment on this? Is anyone able to project for us what this looks like at 3000 and 3500 spell power? My napkin math tells me Renew (when I'm raid buffed, around 2800 sp or so) is worse than Flash Heal most of the time, unless Renew can heal for something like 90% effectively (and many times it's 25% or less).

Anyway, Renew is still bound to my 1 key. I still use it here and there, and usually out of boredom (though it definitely is part of my repertoire on Malygos now). I maintain the problem with using Renew is less about the spell and more about the encounters in WOTLK and the instant-heal raid environment, but efficiency probably needs to be addressed, too.

I'm too lazy to provide you with any real numbers right now, but renew scales better with spell power than flash heal but has a smaller base heal. In other words, no. Renew gets better and better compared to flash heal the higher your spell power is. Relativley speaking of course, renews numbers aren't terribly favorable for what a hot should be even at 3k spell power right now.

Let's take a holy priest. Renew gains 1.88 HP per spell power over the duration (untalented; talents add 15% to the total value); Flash Heal gains (3/7 * 1.88) + 0.2 HP per spell power. Both benefit from Spiritual Healing.

So at 2800 SP and 25% crit (ignoring both SoL and IHC for the moment, and assuming that none of this overheals at all ever):
Renew talented: (1400 + 2800 * 1.88) * 1.1 * 1.15 = 8429.96
Flash Heal noncrit: (2049.5 + 2800 * (3/7 * 1.88 + 0.2)) * 1.1 = 5352.05
so with crits: 5352.05 * 0.75 + 5352.05 * 1.5 * 0.25 = 6021.06

Mana costs for glyphed Flash Heal and Renew are similar unless Flash Heal picks up Serendipity.
Flash Heal + glyph + Serendipity: 625 - 157 = 468 mana, so 6021.06 / 468 = 12.87 HPM.
Renew (without MA): 656 mana, so 8429.96 / 656 = 12.85 HPM.

In other words, I've cut a whole lot of corners here, but Flash Heal is about as efficient as Renew IF you pick up Serendipity (and Rapture has an even stronger effect for a discipline priest). A crit Flash Heal can also give you SoL for more efficiency, IHC for more HPS, or Inspiration for more mitigation on a tank.

Note that Flash Heal gains ~1.01 HP per spell power, so as you gain spell power Renew pulls farther and farther ahead in potential HPS (even counting crit).

But this isn't really a fair analysis. The question is not whether to replace PoM and CoH with Renew; we all know that those spells are fantastic on Sapphiron, and do more HPS/HPM than Renew in any situation where they have full effect. The question is whether to use it instead of Flash Heal, and on those terms it looks like a good option (better heals per cast time, heals per mana just as good, and you can cast it while running around if necessary).

Concerning Flash Heal vs. Renew-
I would like to hear other people's impressions on how many renew ticks they usually obtain in a raid before someone else heals it up. I think I have a better chance of being an arena gladiator than having a renew tick for anything more than a few ticks before someone else heals the deficit. Although, of course, Flash Heals can get sniped as well and land for nothing.