Elitist Jerks WotLK Healing Compendium v3.2: Same Old Thing

 06/23/09, 9:38 AM #571 tedv Observation: I am awesome     Hope Goblin Priest   Mal'Ganis Forgive me for the criticism, but how exactly do you expect someone to take both [Starshine Circle] and [Sunglimmer Cloak] for the holy BiS list when they are alternate rewards from the same quest? On the subject of trinkets, [Eye of the Broodmother] is just hands down better than [Illustration of the Dragon Soul]. The stack is easier to rebuild and it provides more total throughput. [Spark of Hope] beats [Pandora's Plea] by a sizable margin, even for discipline. 100 spirit isn't too much worse than 108 int, and the 42 mana cost reduction helps far more than a random proc that increases your spell power at some moment when you may or may not need additional throughput. If you are including item level 239 hard mode loot on the list, then the best in slot weapon has to be [Val'anyr, Hammer of Ancient Kings]. I suppose a gear list is nice, but it's good to hear justification. Besides, stating that cloth pieces dropping from hard modes are best-in-slot isn't particularly ground shaking, and I suspect that most priests in guilds that are farming hard modes already know what items are best. A discussion between two alternatives at the same item level is the type of thing that generates value. So all that said, can you explain what criteria you used to decide "best" and why it's a good heuristic to use?
 06/23/09, 12:25 PM #572 constantius Soda Popinski   Nidaba Pandaren Priest   Windrunner If you want an intellect regen trinket, [Meteorite Crystal] is hands-down better than [Pandora's Plea]. [Spark of Hope] is more regen than either of them, though. Personally, I'm aiming for Spark of Hope for Holy, and Meteorite Crystal for Disc, since the spirit->spellpower conversion makes Spark of Hope less valuable when I don't have SG. [e] Actually, I would wear Amice. Interesting choice; I had originally discounted it. The net gain (assuming you socket +19 spell, +16 int) is basically 7 spirit, 5 spellpower over T8, but that is a gain. Of course, you lose stamina, which for most hard-modes is actually a consideration. Last edited by constantius : 06/23/09 at 12:31 PM. Anyone who cannot cope with mathematics is not fully human. At best he is a tolerable subhuman who has learned to wear shoes, bathe, and not make messes in the house. - R.A. Heinlein
06/23/09, 2:19 PM   #573
l337n00b
Von Kaiser

Human Priest

Vek'nilash
 Originally Posted by grghrkn This is key. Every model of performance must necessarily be applied against actual raid scenarios. But we shouldn't confuse our individual, anecdotal perceptions of a raid with hard data. If evidence for the value of any given talent exists, we should be able to formulate a model and detect in our logs. At the very least, we should be able to compare performances between those using two different models to move towards an answer. But we need to formulate a model not of Serendipity's weakness, but of it's strength. And then we need to come up with hard data that demonstrates the validity of this model. We're not debating which Jonas Brother is the cutest. We're discussing the value of a specific optimization to a purely mathematical system. We should be able to come up with a model that demonstrates the value of Serendipity and point to data that validates this model rather than merely relying on authoritative pronouncements.
I love to make mathematical models, but I think for a talent like serendipity, anecdotal evidence and thoughtful arguements constructed out of scenarios we have a common experience of make more sense.

People talk about preventing death as being the goal of healing, but how can we tell if Serendipity (or any other talent) prevented a death? We could parse the whole log given people's known max health and figure out exactly whether the heal landed in time to get ahead of lethal damage to anyone. But if it was healing for less, or if it was slower, maybe we would have behaved differently at a different time. Maybe the person we were healing was about to hit a health stone and decided they didn't have to. Maybe we *should* have hit a different heal at a different time but we just weren't good enough to know we had to do it.

How much can we expect from our talent points? If Serendipity is doing it's job, then it is probably going to save your raid from a wipe with an incredibly low frequency. I don't say that to complain about Serendipity, that's just what it's function is. Nature's Swiftness saves raids from wipes very rarely, but everyone agrees they should take it. What if you were bugged only had 68 talent points (and for some reason people brought you anyway). Do you think that suddenly the raid would be wiping all the time to fights it can normally handle? A raid probably brings 5 to 6 healers to any given fight, it is hard to believe that losing 3 of the 71 talent points from one of those healers would make much more than a 1% difference to the chance of a wipe? (If we carefully selected the most relevant and impactful talents it might, but Serendipity is a middle-of-the-pack talent, one that few people would cut but not GS, Soul Warding or Penance)

If doing something changes your chance of wiping by 1% then to get "hard data" of any statistical relevance, you would have to track hundreds of boss attempts both with and without the talent to see if it made a difference. And how would you account for the facts of different raid compositions and gear levels? What if one style ultimately outperforms the other with significantly more practice, but underforms until the first style has plateaued? What if one way of spending talents would make the whole raid better off if only they learned to use it properly but generally is worse because people don't? What if one style is skewed by the fact that it is superior in the general case but unviable for one fight?

There are many problems in this world that are too complicated to answer by making a model, plugging in the data and seeing what comes out. Brains, however, are great problem solving engines that often have to justify (after the fact with anecdotal arguments) conclusions that have been drawn from very complicated processes that are beyond the grasp of the person with the brain. And, yes, it's the same way we would decide which Jonas Brother is the cutest, but brains are the best thing we have to tackle problems of this kind of complexity, and we are going to have to live with that.

An idiot is someone who would rather be treated like an idiot than called an idiot

 06/23/09, 6:40 PM #574 Sinndir Great Tiger     Sindir Night Elf Priest   Medivh Our proposed tier-9 set bonus' are in: Priest T9 Shadow 2P Bonus (Devouring Plague) (Class: Priest) -- Increases the duration of your Devouring Plague spell by 6 sec. Priest T9 Shadow 4P Bonus (Mind Flay) (Class: Priest) -- Increases the critical strike chance of your Mind Flay spell by 5%. Priest T9 Healing 2P Bonus (Prayer of Mending) (Class: Priest) -- Decreases the cooldown on your Prayer of Mending spell by 2 sec. Priest T9 Healing 4P Bonus (Prayer of Healing) (Class: Priest) -- Increases the critical strike chance of your Prayer of Healing spell by 5%. Very blah to me, considering that the PoM spell could bring us to a 5 second PoM with 5 charges, that is a lot of needed bouncing. The 4 piece is 5% less crit than our tier 8 2 piece gives... pretty lack luster.
06/23/09, 7:07 PM   #575
Awina
Von Kaiser

Azjol-Nerub
 Originally Posted by Thedankson You're trying to oversimplify things to prove your point. Empowered Healing also affects Gheal but you decided to leave that out, which I agree with. The problem is that you list Serendipity affecting Gheal, which makes it "watered down". Regardless of what a tooltip says, the only way to judge and evaluate talents is going by how they affect your raids. Serendipity is a PoH talent and EH is a FHeal talent. Gheal should be left out of the equation. If anything it gives a *very* slight edge to Serendipity as having a hasted emergency Gheal > having a more powerful GHeal. I don't see myself giving up Serendipity after the PoH nerf unless they change the boss fights. It's a talent that fits perfectly with a majority of the boss fights in Ulduar. It's more utility than straight out throughput, similar to B&S or Tuskar's.
Sorry, I posted that in haste and forget that people can't see my current spec, which indeed does not include Empowered Healing either. I meant Empowered Renew, and I'll edit that above - my bad.

Obviously the ultimate test is how it performs in raids - nobody is going to ignore that over theory. The talents were balanced (well, you know what I mean) around all the spells they affect and Blizz assumes a certain mix of those so that is what I mean by watered down. That doesn't, in and of itself, make them unworthy.

Either way, for me, in my current raids, I applied the same logic to both Serendipity and EH and deselected both of them. Mileage may vary for others. My % of FH has gone down as my Renews have gone up, thus watering down EH even further. /shrug

 06/23/09, 7:51 PM #576 BobTurkey Glass Joe   Turkelife Blood Elf Priest   Nagrand BIS discussion @glasswizard - I don't believe that quest is repeatable. This is a BIS list, which is intended to inform which is the best item in any given slot. Thanks for Starwatcher's Binding, I hadn't seen that before. Neither of the T8 bonus are especially valuable for Holy, thus Leggings of Profound Darkness is better than Conqueror’s Leggings of Sanctification. @tedv - I don't expect anyone to get all the items of this list, as per my comment @glasswizard above. The stack is easier to build on Eye of the Broodmother, but it is also smaller, which is not compensated by the crit rating IMHO. Spark of Hope is inferiour to Pandora's Plea for Disc. Spark of Hope provides around 130-140 MP5. To be superiour to Pandora's Plea for Disc it would have to provide more like 165-170 MP5. This is mostly because Spirit is quite a weak stat for Disc. For Disc spirit is about one-third as useful as Intellect. Good point re: the legendary mace. Will have to give that some more thought. Criteria is based on my theory crafting (Priest Theorycrafting – Summary (Part 6) � BobTurkey’s WoW Blog). Why is this a good heuristic to use? I've found to to be effective and so far i'm yet to see a convincing argument why it is radicaly wrong. There are not too many other attempts to empirically come up with a better set of numbers and those that do exist either support my theory or recommend weights without providing any indication of how they were arrived at. I'd love some constructive alternative be presented. @constantius - Yes for pure regen Meteorite Crystal is better than Pandora's Plea, however assuming you give any weight to throughput at all then Pandora's Plea is superiour. Details of the calculations are available here: Healing priest trinkets in 3.1 � BobTurkey’s WoW Blog. Spark of Hope is superiour to Meteorite Crystal for mana regen for Holy, but it is not for Disc as I said in my original list. Once again this is because the 100 spirit on Spark of Hope is weak for Disc but fairly useful for Holy. The 111 intellect on Metorite Crystal is very strong for Disc and useful for Holy. If you like i'll post a step by step calculation of this (although much of it is repeasting what in my post linked above). @Sinndir - I agree the T9 bonuses are returning to the relative ho-hum of T7 set bonuses.
06/23/09, 8:25 PM   #577
Raiynne
Glass Joe

Human Priest

Blackrock
 Originally Posted by Sinndir The 4 piece is 5% less crit than our tier 8 2 piece gives... pretty lack luster.
I hear that. Pretty disappointed with the 4 piece holy/disc bonus considering that this will be the third time we've seen the same bonus albeit with different percentages since vanilla. If they want to recycle bonuses I would like to see them bring back things like the 2 piece T5 bonus.

One can hope since the PTR is in it's infancy these set bonuses are placeholder and will be changed in the future.

06/23/09, 8:45 PM   #578
moowalk
Don Flamenco

Troll Priest

Khaz'goroth
 Originally Posted by BobTurkey @glasswizard - I don't believe that quest is repeatable. This is a BIS list, which is intended to inform which is the best item in any given slot. @tedv - I don't expect anyone to get all the items of this list, as per my comment @glasswizard above. The stack is easier to build on Eye of the Broodmother, but it is also smaller, which is not compensated by the crit rating IMHO.
You're being inconsistent then, since you list 4 pieces of tier 8 for discipline, which are only BiS if all 4 pieces are used.

Either it's a BiS complete list, in which case you can include one quest reward and should list 4 pieces of Tier 8.5. Or it's a BiS for each slot, in which case you can include both quest rewards but you shouldn't include the 4 piece bonus.

 Originally Posted by BobTurkey Neither of the T8 bonus are especially valuable for Holy, thus Leggings of Profound Darkness is better than Conqueror’s Leggings of Sanctification.
I disagree about the 2 piece bonus.

 06/24/09, 6:17 AM #579 Havoc12 King Hippo   Shaarra Night Elf Priest   Silvermoon (EU) After a lot of experimentation I have found that the value of int in terms of spirit is grossly overestimated Here is the numbers which fit my situation over many runs: HC uptime: I have no real way of properly tracking this. The common method of estimating uptime from WWS reports is inaccurate. I use a simple simulation to calculate my uptime based on my casting pattern. I never take SoL so that greatly increases my overall uptime. The worst case senario is 4 eligible casts per 12 seconds. This may be anything between 2 and 5 casts depending on the exact pattern and the vast majority of hte time they come grouped up. Thus assuming an average of 4x1.3sec intervals and 4 possible procs, with a 28% crit rate this gives me an uptime of ~43%. However this is not the case for the entire fight. I estimate that 5% in any fight is non casting time i.e. a period of 15+seconds where I dont cast anything. This is 100% ooFSR time and 0% HC uptime. The rest is split between 95% iFSR with 43% HC uptime and 5% ooFSR time with 90%+ HC uptime. This gives me an overall factor for FSR and HC of ~0.667. My raid buffed spirit and intellect are ~1400. I am not including the use of the darkmoon greatness card, which raises my overall spirit by about 100. Regen from spirit = R $R_{spirit}= 1.20613*k*f_{FSR+HC}*spi*\sqrt{int}$ At my values this is 704.6810782 total mp5. Adding 1 spirit raises this to 705.1844218 mp5, which is an increase of 0.503343627 mp5 Adding 1 int raises this to 704.9327051 mp5 which is an increase of 0.251626888 mp5 In an 8 minute fight it is not possible to cast HoH and fiend more than once without losing out overall and they are best used chained when your mana is lowish (Remember that the 20% max mana from HoH actually fills up your mana bar, so you gain say 4k max mana and 4k actual mana, making it easy to lose mana if you use it early). Chaining fiend with HoH increases fiend effectiveness by 16% raising its value to 58%. Getting 4 ticks of HoH gives you back 13.8% total mana for a total of 71.8% max mana once for an 8 minute fight. Thus per point of int: HoH+fiend = 0.718*16.5*5/8/60 = 0.12340625 mp5 Mana pool = 16.5/8/60*5 = 0.171875 mp5 Replenishment (95% uptime) = 0.1959375 soon to be nerfed to 0.15675 Overall values: 1 spi: 0.503343627 1 int: 0.742845638 soon to be nerfd to 0.703658138 This is a far cry from the 100% value that is normally being thrown around, at least for the circumstances I described, which are very accurate for my case. Even for people who take SoL, which can decrease spirit regen by as much 5-10%, its nowhere near 2x for intellect. Considering that spirit also gives spellpower, I would definately say that spirit and int are pretty close in overall value. An important change in terms of regeneration in the new patch is that mp5 has apparently had its ilvl value decreased by 25%(at least on items, don't know if its the same on gems).  Misread patch notes its 25% increase  ================================================================= New patch. I do not understand why ppl are tripping over the nerf to PoH. Its no big deal. The HPS of PoH is currently so far above incoming DPS that you can nerf it even further without affecting the ability of priests to heal any encounter. All that the nerf is going to do, is decrease the position that priests have in the healing meters and probably not by much. PoH is like 35-45% of our healing. Blizzard has clearly moved away from the HPS limited aoe healing design. AoE healing is now time limited. That means the crucial factor in determining aoe healing success is how many targets you can hit with a 3k+ heal within a given window of opportunity, that generally lasts less than 10 seconds. Although this heal might not save them on its own it will almost certainly guarantee that they will live long enough for another heal to save them. If you hit anyone ONCE with PoH, nerf or not, they can be effectively considered saved for the immediate present. I am quite surprised to hear ppl arguing about blessed resilience over serendipity. Blessed resilience is a talent that has no value other than a non-visible increase in healing meter output. It does not help you heal more ppl within the window of opportunity and its so small it does nothing to help with underhealing, which is already a very minor issue with aoe healing. Serendipity on the other hand can increase the number of heals you can land quite significantly, by allowing a different spell sequence: No serendipity Precast PoH (it lands after 1.5-2 seconds). PoM/CoH 4.5-5 seconds PoH 7.5-8 seconds. If you do get another CoH at 8 seconds you will sacrifice more PoM ticks, if you maximise PoM then the second CoH is going to land at 9+ seconds Alternative: Precast PoM, and CoH when the damage hits (1.5 seconds and you have a PoM jump already). Hasted PoH, (3.5 seconds) PoH+CoH (6.5 seconds) The difference between the two solutions is vast. In the first one you deliver a quick heal, but then there is a lag phase, while you stick PoM on and wait out the GCD and long cast time of the second PoH. With the second solution you have a PoM ticking already and deliver an instant CoH and a hasted PoH, which will land relatively early after the burst begins (3.5 seconds) and you have time to land a second big fat PoH, CoH combo while it still matters. Even better you also have the option of adding in a binding heal+PoH combo if you needed it, which only takes 3.5 instead of 3 for a single unhasted PoH. Blessed resilience is a talent whose only function is to raise your position in the healing meters. Serendipity is a core talent that greatly boosts the your burst aoe HPS. In my view there are only four talents in the entire holy tree that are mandatory: Serendipity, GS, divine providence and spiritual healing. A lack of any of these 4 talents results automatically in a significantly inferior holy build. The remaining talent choices are very flexible and a matter of playstyle. Blessed resilience is a talent of no consequence. Last edited by Havoc12 : 06/24/09 at 7:32 AM.
06/24/09, 6:41 AM   #580
Ellyh
Don Flamenco

Human Priest

Hyjal
Havoc, where do you get the 50% increase in MP5 you refer to? is it on the PTR realm? The reason I ask is that the posted comment from blizzard was for a ~25% increase.

 Mana Regeneration: All items that provide "X mana per five seconds" have had the amount of mana they regenerate increased by approximately 25%.

 06/24/09, 7:07 AM #581 Havoc12 King Hippo   Shaarra Night Elf Priest   Silvermoon (EU) My mistake, I misread patchnotes. I amended.
06/24/09, 12:34 PM   #582
Elimbras
Don Flamenco

Dwarf Priest

Eitrigg (EU)
 Originally Posted by Havoc12 HC uptime: I have no real way of properly tracking this. The common method of estimating uptime from WWS reports is inaccurate. I use a simple simulation to calculate my uptime based on my casting pattern. I never take SoL so that greatly increases my overall uptime. The worst case senario is 4 eligible casts per 12 seconds. This may be anything between 2 and 5 casts depending on the exact pattern and the vast majority of hte time they come grouped up. Thus assuming an average of 4x1.3sec intervals and 4 possible procs, with a 28% crit rate this gives me an uptime of ~43%. However this is not the case for the entire fight. I estimate that 5% in any fight is non casting time i.e. a period of 15+seconds where I dont cast anything. This is 100% ooFSR time and 0% HC uptime. The rest is split between 95% iFSR with 43% HC uptime and 5% ooFSR time with 90%+ HC uptime. This gives me an overall factor for FSR and HC of ~0.667.
I have difficulties to understand your model and assumptions.
If I try to rewrite what I understands, it reads as follows:
2. When casting, you cast about 2 "HC eligible" spells per 8 sec.
3. 5% of the time is standing still, which means ooFSR and without HC
4. 95% 95% of the time (ie 90.25%), you assume being iFSR, and with 2 "HC eligible" cast in 8s, you get 48% HC uptime
5. For the rest of the time (5% * 95%, ie 4.75%), you assume being ooFSR, with 90% (or plus) HC uptime ?

What questions me the most is the 4th assumption: why do you assume such a high HC uptime when ooFSR ? It would mean you actively take beaks after HC procs, and you're able to do it (ie., no healing is needed right-now). It might be the case, but I like to have it clearly stated, and if possible, explained and argumented.

06/24/09, 1:33 PM   #583
Havoc12
King Hippo

Night Elf Priest

Silvermoon (EU)
 Originally Posted by Elimbras I have difficulties to understand your model and assumptions. If I try to rewrite what I understands, it reads as follows: Your crit rate is about 28% When casting, you cast about 2 "HC eligible" spells per 8 sec. 5% of the time is standing still, which means ooFSR and without HC 95% 95% of the time (ie 90.25%), you assume being iFSR, and with 2 "HC eligible" cast in 8s, you get 48% HC uptime For the rest of the time (5% * 95%, ie 4.75%), you assume being ooFSR, with 90% (or plus) HC uptime ? What questions me the most is the 4th assumption: why do you assume such a high HC uptime when ooFSR ? It would mean you actively take beaks after HC procs, and you're able to do it (ie., no healing is needed right-now). It might be the case, but I like to have it clearly stated, and if possible, explained and argumented.
1) Yes my crit rate is 28%, its too high for my liking, but I dont have a wide choice of gear. Though other healers might have less or more crit rates, the results are roughly the same as long as they have a crit rate of 25%+

2) My casting pattern gives me an average of 4 HC procing casts per 12 seconds (not 2 per 8 seconds). Notice I say casts not spells, because binding heal is one spell but it has 2 casts.

3) Yes. 5% of the time I have to take a long enforced break. For example like in between the phases at mimiron. Its rough, but it does not need to be exact. Its in this ballpark.

4) That is correct 90.25% of the time I am iFSR and have an HC uptime of 43% with this casting pattern.

5) I do not actively take breaks during HC, I simply choose not to take a break UNLESS I have HC. That means if an opportunity for a break arises I will not take it unless HC is up. If HC is not up I will continue casting until it IS up and then see if I can take a break.

 06/24/09, 3:06 PM #584 binaryz Glass Joe     Lime Blood Elf Priest   Thrall I don't see this being mentioned but the Glyph of Holy Nova was changed on the PTR. It now increases damage & healing by 20%.
 06/24/09, 3:19 PM #585 constantius Soda Popinski   Nidaba Pandaren Priest   Windrunner Guess they really don't want us using it as a CoH substitute. Oh well. Anyone who cannot cope with mathematics is not fully human. At best he is a tolerable subhuman who has learned to wear shoes, bathe, and not make messes in the house. - R.A. Heinlein

 Elitist Jerks WotLK Healing Compendium v3.2: Same Old Thing