Every spell will cast more quickly with more Haste, up to a 1 second GCD. Haste does not reduce CDs, though. MoP, more I think than any other expansion, has huge disparities in spell efficiency. Looking at several of the top-ranking logs on some heroic encounters this tier, it looks like about 50% of a given Disc. priest's heals come from these efficient spells with CDs. Having more Haste doesn't give you more of these efficient spells - it gives you more of the spells that you can spam, which (with the arguable exception of PoH) are not nearly as good.

Havoc, please correct me if I'm wrong, but:

The expected value of a heal is a factor of 1 + c(2+m) times its base heal. In your example, comparing approximately 16% Crit to approximately 12% Crit, your expected heals increase by a factor of (1+.16*2.475)/(1+.12*2.475) = 1.076.

You make a good point about haste helping to squeeze things in before a window closes, but this can also go the other way, if you don't have enough haste to get that extra spell in.

Looking at the log you linked, I see PoH with a crit/noncrit healing ratio of 2.50, Cascade with a ratio of 1.79, Penance with a ratio of 2.06, Atonement with a ratio of 1.91, and PoM with a ratio of 2.60. Inner Focus, used only when people are low, will certainly increase the apparent effectiveness of Crit. It also devalues Crit, because with it up you'll crit regardless of your Crit rating. It's also defensible if not ideal to macro Inner Focus to PoH (I do), in which case you won't see the kind of bias that you're talking about.

Why in the world would 60% of your healing be overheal? That's absolutely atrocious.

You forget that crit attonement leaves aegis on target, which is hardly will be an overheal.
And crit is much useful for Disc in 10 man, because SS will be capped on 2 groups in 2 cast no matter how mastery you have in 500+ gear. In 25 man there are 5 groups, so you will place like 70% of SS cap with single PoH cast, and after you place such bubbles on 5 groups, 1st one will fade completely or partially.

In 10 man, HALF of the raid are nodpsers. If you raise dps of that half, your raid dps will raise significantly. If you can dps as healer in 10 man - thats awesome. And you dont need mastery reforge because you have only 2 groups and with high Int and Crit you will get not so much lower SS, but you have more dps stats and more to raid contribution.

In my 10 raid our healers make as much dps as possible. Because it really matters.

yes, gcd is reduced.

but it is difficult to evaluate the improvement due to the haste on spells which are instant and include a cd.

For example, on a spell which is casted like PoH , you can write that if you increased your haste by 10%, you will increased your Hps by 10% if you chain heals.

You cannot write same think for instant cast with cd because you will only improved the gcd but not the cd.
So you will not able to do more HpS on cascade or PoM because you will not be able to cast more cascade or PoM with haste.
You will only be able to chain an other spell more quickly
The improvement due to the haste will be lower (and pretty lower) than spells like PoH, GH, etc....



i have some difficulty to follow all demonstrations done but with mine, i do not agree to go for haste/spirit for 5.2
We have to take into account what will be the background gameplay of the disc.

Is there is spamming for attonement ? ==>crit/haste (spirit seems to be useless at high level thanks to solace)
Is there a spamming for absorb prevention including PoH without SS ? ==>mastery still better i think because of high overheal.

I see what you mean. What you are seeing can only be two things: 1) An artifact caused by the use of healing cooldowns and temporary raid crit buffs during active healing stages. These don't make crit better they make it worse. Intellect buffs contribute a tiny tiny increase in spell damage 2) Lack of statistical power due to the small sample size and low crit rates. I.e. what you are seeing is just random variation (lots of lucky crits on low health targets) and is not actually statistically significant, meaning it should disappear upon increasing sample size.

Look at this log for example: Details for Dailybreeze - 26-12 19:41 - Unholy Trinity - World of Logs

And this log: Details for Izumî - 02-01 20:54 - Blow - World of Logs

Looking at the top parses is biased because these selected for high HPS, they are enriched for those lucky cases where crit helped more than it should. Looking at lots of different logs all over the spectrum you clearly see that the only spell with consistently high crit/heal ratio is PoH and we know that this is bad news for crit, because it means increased contribution of inner focus to low overhealing crits.

100th log: Details for Trinitysin - 13-12 00:12 - Oto Logs - World of Logs

150th log: Details for Vintoran - 16-12 18:47 - Predicted - World of Logs

200th log: Details for Vintoran - 16-12 18:47 - Predicted - World of Logs

224th log: Details for Evilkyro - 19-12 20:28 - Luce - World of Logs

(225 has divine star) and there is no 250th.

I selected these essentially at random. Looking through the briefly will show you that only PoH has a consistently high avg crit to heal ratio.

However the most important thing to realise and what I tried to explain in my previous posts is that the avg crit to heal ratio is not linearly dependent on the actual contribution from crits. Just because average crit is twice as high as the average heal it does not mean that crits contribute 100%. This is because only the top half the heal is the contribution from the crit. The presence of low and high overhealing phases, lucky streaks with low overhealing crits and the use of cooldowns in low overhealing phases, artifically raise the average crit heal, much more than they raise the actual contribution of crit.

I will show you some examples:

Look at this for example from my own logs when I know I had a cooldown

Details for Shaarra - 02-01 22:42 - After Hours - World of Logs

check this segment: Details for Shaarra - 02-01 22:42 - After Hours - World of Logs

and look at the log for cascade:

[22:59:07.697] Shaarra casts Cascade on Burgenpils
[22:59:08.749] Shaarra Cascade Burgenpils +47435
[22:59:10.355] Shaarra Cascade Whiteredbull +42880 (O: 9276)
[22:59:10.355] Shaarra Cascade Sèraphim +51995 (O: 161)
[22:59:11.586] Shaarra Cascade Nerad +24441 (O: 21457)
[22:59:11.586] Shaarra Cascade Doomguard +*43164* (O: 42789)
[22:59:11.586] Shaarra Cascade Undu +41725
[22:59:11.586] Shaarra Cascade Naghtorin +41725
[22:59:12.773] Shaarra Cascade Fantomen +0 (O: 41725)
[22:59:12.773] Shaarra Cascade Kérberos +0 (O: 41725)
[22:59:13.148] Shaarra Cascade Mindbender +0 (O: 41725)
[22:59:13.148] Shaarra Cascade Risen Ally +0 (O: 41725)
[22:59:13.148] Shaarra Cascade Bõllex +*52783* (O: 33171)
[22:59:13.148] Shaarra Cascade Ashghost +0 (O: 45898)
[22:59:35.262] Shaarra casts Cascade on Undu
[22:59:35.723] Shaarra Cascade Undu +27014
[22:59:37.333] Shaarra Cascade Whiteredbull +42880 (O: 4025)
[22:59:37.333] Shaarra Cascade Rinotzk +4118 (O: 42786)
[22:59:38.152] Shaarra Cascade Nylaki +40179
[22:59:38.394] Shaarra Cascade Shaarra +46904
[22:59:38.545] Shaarra Cascade Naghtorin +46905
[22:59:38.948] Shaarra Cascade Nerad +51595
[22:59:39.353] Shaarra Cascade Bõllex +28953 (O: 12841)
[22:59:39.353] Shaarra Cascade Kérberos +0 (O: 42401)
[22:59:40.180] Shaarra Cascade Sèraphim +0 (O: 46905)
[22:59:40.180] Shaarra Cascade Meätball +39177 (O: 7728)
[22:59:40.180] Shaarra Cascade Gosak +0 (O: 46904)
[22:59:40.180] Shaarra Cascade Biermeister +36834 (O: 10071)
[22:59:40.180] Shaarra Cascade Kurrelgyre +0 (O: 46905)
[22:59:40.180] Shaarra Cascade Elooss +46905

How lucky was that? None of the crits overhealed completely even though most of the normal heals did, but what is the actual benefit of crit here: Only one crit actually had any benefit and that was for about 11k. So despite crits contributing 95947k to my total healing only 11k of that was the actual benefit of crit heals over normal heals.

7% of the heals are crits but the benefit of those crits to overall healing is just 1.5%. Nevertheless the average heal to average crit ratio is 1.9 to 1!!

If you looked at the average crit to heal ratio you might have thought that crit was nearly at 100% efficiency, but in fact it didn't even reach 20% efficiency. If I had zero crit I would not notice any change in my healing.

The contribution of smart heals, inner focus and cooldown usage during low healing phases, artificially raises crit averages, but that does not mean criticals contribute more.

To give you another example I return to the rank 3 report from 25hc vizier in HoF and specifically this segment

Details for Th�*llo - 11-12 15:33 - Ellui - World of Logs

Crit healing total is 154671 but only 49 200 was the actual benefit of crit. 15.6% of the heals were crits but the actual benefit of crit was just 7.5% of the total. That is 56% of the benefit we expect for this crit rate. Ratio of avg crits to heals 1.670590636908

Lets look at a high HPS segment

Details for Th�*llo - 11-12 15:33 - Ellui - World of Logs

1 out of 9 crits hit a low deficit target, while 7 out of 27 normal heals hit a low deficit target. This is a fairly lucky event for crit.

The average crit to average heal ratio is 2.7 to 1

crtis contribute 261778, which is 48.96% of total crit heals and 23% of the total healing.

25% of the heals are crits, so the value of the crits is 115% of maximal, but the ratio of crits to heals is 2.7:1, when given the actual value of crits we should expect roughly 2:1

Now lets have a look at the end of the fight, another high HPS phase:

Details for Th�*llo - 11-12 15:33 - Ellui - World of Logs

125288 contribution from crits, which is roughly 30% of the total crit healing and 8.8% of the total healing. 22% of the heals are crits so we expect 18% contribution, so the value of crit is just 49% its maximal value. The avg crit to heal ratio is 1.5:1

Thus the relationship between the average crit to heal ratio is very difficult to predict, because the system has several variables it is difficult to account for the full variance without fairly humongous numbers. So looking at the average crit to heal ratio does not necessarily tell you what the actual value of crit is even if the ratio is 2:1. It is basically a very bad way of determining crit value. Averaging things is always dangerous as it eliminates variance information, which is pretty important in determining whether your results are accurate.

Looking at the average overheal is a good way to find if crit is useful or not. High OH values automatically mean low contribution from crit except in extreme cases where the vast majority of your heals either overheal completely or do not overheal at all.

In general you can expect crit to have a 20-50% lower value than the maximum.

I was curious about the Spirit Shell change on PTR, did a little testing.

First, hypotheses: If Spirit Shell is to correctly mimic the average throughput of the non-SS version of each spell:

PoH SS = MeanPoH*(1+C)*(1+0.5*(1+M)), where C and M are crit% and mastery% from charsheet
Other SS = MeanHeal*((1-C) + C*2*(1+0.5*(1+M))), carefully accounting for non-crit and crit separately, but can be reduced to
MeanHeal*(1 + C*(2+M))

On my PTR premade, with proc items and crit meta hat removed, spellpower is 16952, C is 8.63%, and M is 36.4%. Mean PoH in the spell pane is 22,893 and mean GHeal is 59,850.

Predicted PoH SS is 22893*(1+0.0863)*(1+0.5*1.3640) = 41829. Actual is 41831. Good.
Predicted GHeal SS is 59850*(1+0.0863*(2+0.3640)) = 72060. Actual is 72059. Good.

------

The formula works so closely actually, that we can now put a crit meta on and see that it tried to take it into account somehow, but doesn't quite do it correctly. With a meta, crit bonus is 206%, so our formulas become:

PoH SS = MeanPoH*(1+C*1.06)*(1+0.5*(1+M))
Other SS = MeanHeal*((1-C) + C*2.06*(1+0.5*(1+M))) = MeanHeal*(1 + C*(2.09+1.03*M))

With my crit meta hat back on, SP is 18,144. C is 9.1% and M is 36.4%. Mean PoH is 23,892 and Mean GHeal is 62,461.

Predicted PoH SS = 23892*(1+0.0910*1.06)*(1+0.5*1.3640) = 44063. Actual is 43934. Without the crit bonus, it would be 43843.
Predicted GHeal SS = 62461*(1+0.091*(2.09+1.03*0.3640)) = 76471. Actual is 76129. Without the crit bonus, it would be 75898.

So the simple prediction works quite exactly without the crit meta. With the meta, it's doing something to account for it, but isn't doing so quite correctly--the added amount is smaller than it should be.

Havoc12, and again you forget about DA, that critheals produce.

Its not just Heal*2, most of it should be OH and thats not a problem.
Critheals for Disc is Heal*2 + 0.5*(1+%Mastery)*2*Heal and 2nd part of it is awesome bubble which is rarely overheal.

It is work even better with attonement. It always heal lowest target and shield that targets after crit attonement and thats a little help in random death prevention.

We made almost all boss in heroic mode with 2 healers (exceptions are Empress and first 3 boss in TES) and I spent most of the time on Smite/Penance

Ok, so some player spam heals even when the raid is full hp and gets massive over heal. They also happened to get a good parse rank because they ninjad heals from their co-healers. That must be the best way to heal. Nothing to do with poor spell choice or running with weak or few co-healers.

In my last log my over heals on average are about 35%.
Stone guard 16% overheal (2 healing, with 3 healers its probably higher)
Feng 27% (low damage phases probably made me overheal with atonement a bit)
Garajal 29% (intentional overhealing and flash heal spam)
Kings 38% (very very low damage for the majority of the fight, probably solo-healable)
Elegon 35% (cheesy mechanic)

Overall my poh overheal was 40% (i consider that very high, i can probably do better)
SG 40%
Feng 51%
Gara (did not poh even once)
Kings (no poh except during ss for maddening shout)
Elegon 43%

And before you ask, people had to leave on emperors.
This advice you guys give is so bad... you spam poh on full hp groups and then wonder why atonement smart heals are only a small portion of your healing. Try using atonement instead of overheal spam.

Even without looking at the exact numbers I feel Haste is superior to Crit in a very intuitive way.
There's no argument against a crit being really powerful and yea, more of those is of course better. But crits are still random. I don't like random as a healer. In fact I never like random when it comes to keeping people alive.

I think i'll be sticking with a spirit >>> haste >= mastery build, healthily balancing the haste and mastery (really just depending on how much haste I can support with my spirit). Haste will -always- help me keep people up when the big damage comes, and will attain much higher "spike" healing in an on-demand fashion. Using Borrowed time as a multiplier for this is even better. Mastery also is (still) good for spirit shell and Prayer of Healing.
And albeit the mastery is pretty useless for single target healing, I don't think Crit is any better for that (seriously, how many times has it happened that Flash Heal hits a target that a shaman or Paladin already covered with their semi-instant cast big heals? especially in 2-heal teams on 10man.) We generally don't have a lot of reactive healing spells, it's not how our class works either. If someone's about to die we shield them and let the other healer throw in the heal 1-2 seconds later, or do a big heal ourselves if needed. We don't just start casting our flash heal like paladins or shamans would. (and in a 2-man team I really feel these 2 are our best healing partners)

And if you want to drag atonement in there, even then Haste is a lot better than crit. Haste is the most effective way of increasing your healing (and damage!) done with a penance > HF > smite rotation - And more output with that means more periods where you can effectively heal the incoming damage with just that and that will save you mana in the end, make evangelism stacking easier and quicker and both of those are things you can redirect to the periods of heavy incoming damage.

All of that works against relying on crits and favoring "stable" stats that will do what you need them to do, when you need them to do it. At least that's my perspective of how a disc priest is at it's best.

-----

And to comment PerkeyOne - Yes, the top parses are all heal-sniping aegis-stackers. And to an extent that IS an effective healing strategy. But only to a certain extent. As long as your mana allows it it's a good thing to throw in extra PoH to get aegis stacking (and maintained). For example before wind lord's rain of blades in the last phase, it hits so hard you just don't want to take it on with the couple very slow PoH's that you can cast -while- you take damage. You stack up aegis (if not spirit shell) beforehand to smoothen the damage out and it will make things a lot easier for you and the other healer.

But yes, in top parses it gets absolutely abused to get more HPS in. I'm not blaming them, it's fun

Sniping heals by absorbs = damage prevented. There is a reason why disc priests from top ranking guilds heal like this. Your job as a healer is to do maximal HPS and to prevent people from dying. Spamming absorbs is your job as a disc priest. You seem to think that its good to fart around waiting for a big spike to hit before you heal so you don't overheal, but that is not right. By building a big absorb stack you can greatly reduce the damage people taking from big spikes and dramatically reduce their chance of death.

Its not bad advice its sage advice. What possible benefit could I get from spamming atonement when the raid is high health. Lots of overheal and low HPS. A PoH with 100% overheall still leaves the equivalent of 40% in aegis, which you can maintain and stack. Its not abuse and its not inefficient, its the most effective way to use your class.

My mana with 495 ilvl now is good enough to spam PWS/PoH/Cascade non stop. Is there any reason to not do so?

Atonement is a low proportion of healing in high parses because it is a low HPS healing method.

Dont forget the blade throw on windlord. In 25 man that thing often hits the melee group like a track easily 80% of their HP in one hit and frequently it double hits them. Without aegis stacking you get significantly more deaths.

Strange.
you are not in agreement because you do not speak about the same gameplay.
One is talking about bubble spam, the others are talking about a less agressive PoH spamming and more attonement healing.

It is not a surprised that your conclusion is not the same.


Havroc, if i have well understand your comments,

1. on 5.2, you will run for haste with the gameplay that you have described (buble spam with PoH to prevent damage with and without SS).
2. you were running mastery on 5.1 with the same gameplay.

Change of your priority is due to the modification of the mastery effect on SS because on 5.2 , SS will "only" have the same behavior as PoH in term of stat except that critical and intel will not create overheal of this spell since it is converted into absorb too.

I am surprised of your conclusion because if we evaluate the impact on SS, for sure, you have found that SS is reduced by 1 - (1.5 + 0.5%M)/(1.3 + 1.3%M)
With %M=50% for example, the nerf is around 10% only on SS POH.

Conclusion :
You are changing priority because only 1 spell has been nerf by 10% but you still run for POH spamming with high OverHeal value when SS will not be used on PoH.

If you go for haste instead of mastery, you willl have the benefit to win 1 cast under SS.
It is good for 25 men raid but not necessary so usefull with 10 men raid.

I agree that to win 1 cast under SS is a good point but there is a breakpoint where haste is not necessary anymore even if it helps also to continue to spam PoH more quickly after SS activation.

The problem is that without SS, your haste will do exactly what it does actually.
it will create OverHeal at a high value (50 or 60%, and maybe more) and your Aegis done by PoH spamming to prevent or absorb damage will be lower than the one done with a mastery orientation.

Of course, i do not know yet the profil of new raid and if there will be some sustained aoe which will reduce the overHeal of heal part of POH but if it is the same profil as it is on 5.1, haste might be converted mainly into OverHeal like it does today.


Also, running for haste has 2 additional impacts :
1. You must chain cast ==> will it be possible to dot it on new raid profil ?
any stop of casting and the haste benefit's collapses

2. More Haste need more Spirit and your choice to choose Spi/Haste is consistent.
how much do you evaluate to add spirit because you will choose a haste priority ?
how much these points will improve the HpS if they are used into mastery instead of Spirit ?

By taking into account a haste priority, you do not have the right to compare the performance of X points of haste with X points of mastery because Y additional points of spirit will be spend in haste orientation.



I want to be clear, i do not say that you are wrong but i think that it is not so black & white as you said.





Last point :
I have heard that spirit on gear will not be easy to have on 5.2
If it is the case, attonement might be the real background gameplay (when no high damage must be prevented or healed) instead of buble spamming to build absorb because helps to solace modification, the dps spamming for attonement will not cost mana even with a low level of spirit.
(i have tried on PTR that with 6500 spirit , i can spam attonement without mana reduction)

Due to this background gameplay, i think haste will be a good stat with the critical but not with the bubble spam if the OverHeal part stays at the same level as it is today.



Agree on that with metagem.
maybe, it is necessary to create a change request.