PvE HPally Theorycrafting Pt.2

Prev 1 2 3 10 Next
I would really like for a paladin that uses glyph of beacon of light to post some logs of their raids. I find it really hard to imagine situations where beacon is swapped enough to make it a useful glyph.
Each swap costs you 6% base mana, or about 3/5ths of a Holy Light cast, so its not difficult to do the math on that. At best you only do this once (refresh, not swap necessarily) and at worst this could be many swaps done each time a tank switch happens (Chimaeron, Halfus, or Magmaw perhaps).

Personal preference, I think. I'm guessing one is better off using another glyph in a majority of encounters.

To beat Glyph of Divinity (which I assume you're hating on me for not having) it would need 9500-11,000 mana spent, or 19,000-22,000 (it is still bugged correct?), which is like... a lot of Beacon swaps. 8-16 swaps to match it, more to beat it, unless you manage to use LoH twice in one fight.

Sorry I think this was all mathed previously in the thread but it may be buried. :D
Personally, I like to run with just 1 point in TOR and accept the occasional free HP as a bonus, not planning for it as a guarantee.
At the moment I'm experimenting with a setup that forgoes LoD and has a full 3 point TOR, and the result have been somewhat promising. I've had to change my healing strategy significantly, obviously, but in the last couple of raid attempts I've been on I've found my overhealing to be dramatically reduced.
On the other hand, my actual effective HPS hasn't increased much at all.
I'm planning to go back to my usual spec, a single point in TOR and pick up LoD.

By the way, what's the word on the 4.2 change to healing crits and how it effects our stat priorities? My initial math seems to indicate that crit would be more desirable than haste, since it increases our throughput by the same amount without increasing mana costs, but at the same time going full crit would leave us with interminably long cast times that might not keep up with boss damage.
At present I'm thinking keeping haste and crit balanced would be the solution, am I heading in the right direction?
04/29/2011 07:57 PMPosted by Strawbrrymlk
I would really like for a paladin that uses glyph of beacon of light to post some logs of their raids. I find it really hard to imagine situations where beacon is swapped enough to make it a useful glyph.

I would wager that pallies running Glyph of Beacon haven't really internalized just how cheap Beacon of Light has become. It used to be a really expensive spell. But now the cost is relatively trivial.
Actually, I have a question. I haven't kept up with the details of Holy Paladin theorycraft. Why is it that the accepted wisdom is that Tower of Radiance is not worth it?

To me, it seems like you can treat the situation as either getting 50% of the Divine Light through Beacon transfer or getting an extra 1-pt Light of Dawn.

50% of DL = 5866.5 + 0.5765*SP

1-pt LoD (glyphed):
Raid = 3837 + 0.792 * SP
Beacon = 1918 + 0.396 * SP
Total LoD = 5755 + 1.188 *SP

It seems to me that the 1-pt Light of Dawn will end up doing more healing than the Beacon portion of a DL.
05/14/2011 08:02 PMPosted by Coriel
It seems to me that the 1-pt Light of Dawn will end up doing more healing than the Beacon portion of a DL.
I don't know where you got your numbers, but checking the same question with my own gives an obvious answer of "Yes", that is a correct assumption.

1 HP is greater healing output than Beacon healing (especially when you realize the Holy Power heal ALSO outputs more Beacon healing).

Its not that simple though. You also have to account for the fact that the second heal adds time. If you account for that you'll see that the HPS is about 30% higher with the single Divine Light and Beacon transfer than the Divine Light + 1 HP Light of Dawn.

Now, can I find a better way to do this? Yes.. yes I can! Lets try the same thing with a 3 HP Light of Dawn vs 3 Divine Lights with Beacon healing.

Result: Almost identical HPS. One simply takes longer to "rotate" than the other, which is not significant unless specific situation requires different tactics.

Answer: Tower of Radiance is very worth it. If you manage to never, ever land a heal on your Beacon target then you can live without it, but if you ever do, you want 3 points in ToR in my opinion. It opens up valuable options for healing tactics that otherwise wouldn't happen. Heals are needed where they are needed, you can't control that. What you can control is whether that results in more effective healing or not - in this case, you want ToR to make your occasional On-Beacon heal effective.
I might be back, but if we can get a new math nerd, that means I can be lazy!
So here's a first look at what the changes 4.2 will be bringing to Holy Paladins. Please remember this is purely speculative until 4.2 goes live.

I'll first be taking a stab stat weighting, particularly gemming.

For baseline purposes, I'm going to be using a "typical" paladin in mostly ilvl359 gear that will be starting Firelands.

This "typical" paladin has raid buffed:
2000 Spirit
5500 Intellect
8000 Spell Power

In my original theorycrafting, when considering mastery, I simplified its effect by considering a "deprecation" factor of 0.3. I found that 50% of a paladin's healing would proc mastery, and of the shields that procced, 60% were consumed. With the new changes, the 50% eligibility will remain unchanged, but the 60% consumption will likely increase. However, this is speculative until we get some logs published.

What I didn't consider however, was that the other stats have deprecation factors as well, namely overheal. So for comparative purposes, I'm going to deprecate everything else by 20% overheal. For clarity, I define over heal as:

Suppose a paladin heals for 100. 80 is useful. 20 is wasted. This is 20% overheal.

In general, I'm going to assume 20% overheal is pretty much unavoidable in real raid situations.

Adding 10 Intellect
Actual Int - 11.025 (BoK+Armor mastery)
Actual SP - 12.85515 (6% SP raid buff and 10% SP raid buff. I multiplied them which may be an invalid assumption)
Mana - 165.375

10 Intellect will increase spells by:
HS 5.786889377
HL 7.380599143
DL 19.69326957
FoL 14.73899329
LoD 2.253602152 (Per Holy Power Per person)
WoG 3.569880169 (Per Holy Power)
HR 1.144547992 (Per Person Per tick)

Adds the following for regen and longevity (in mp5)
IC Spirit Regen - 1.12759927
Glyphed Divine Plea on CD - 1.2403125
Replenishment - 0.826875

Given the original stats, this is roughly a 0.071% increase across the board. FoL and LoD have larger 0.08% and 0.1% increases respectively. If we consider overheal

Adding 10 Crit Rating
Crit heals now heal for 2x instead of 1.5x. This improves crit. Conviction already basically has 100% uptime, so I'm only going to consider the throughput bonus of crit.

Throughput Bonus
Effective % (20% Overheal)

Adding 10 Mastery
Throughput Bonus

For purely speculative purposes, I'm going to give mastery a 0.5 deprecation factor. This assumes 1/2 of heals are eligible for mastery, but they're magically all consumed.

Effective %

Adding 10 Haste
10 Haste
HPS Bonus
Effective %

Adding 10 Spirit
Given the following stats, adding 1 spirit will add this much mp5.

Estimate by multiplying by 10
bonus mp5
So what does this all mean? Now things get futzy.

The easy stuff first. Crit is better than mastery.

When comparing crit and int, we're seeing how int is better than crit, which is expected. Int has more throughput than crit and does other delicious stuff too. However, from a purely throughput perspective, sacrificing 10 int for 20 crit would be a good thing. But this doesn't consider the mana bonuses of Int, which frankly I haven't found a good way to objectively measure.

My initial recommendation would be to sacrifice 10 int for 30 crit for gemming.

Haste and Int's comparative value is unchanged. At this point in gearing, they're giving about 1:1 in terms of HPS, but again, int gives HPM bonuses and whatnot.

I'd continue to recommend a 10 int for 30 haste tradeoff for gemming.

For mastery, don't trade int.

Spirit is a weird beast. Even if you never use Divine Plea, it takes 194 seconds for 10 spirit to give the same effective mana as 10 intellect. Most fights are 6-8 minutes, so 10 spirit will roughly give 2x the mana of 10 intellect. However, mana<->throughput comparisons are pretty sketchy.

If you're having mana problems, I'd recommend a 10 int vs. 20 spirit tradeoff, otherwise a 10:30 tradeoff.

The tradeoffs are pretty sketchy, so if someone can give good reasoning to change them, I'm all ears.
So I'm going back into trying to determine stat weighting. I don't believe I'm going to succeed, but here's a stab at it. For simplification purposes, I'm going to consider straight DL spam with 0 haste, 0 crit, and 0 mastery. I'm also going to ignore beacon for now.

Again, this is a highly idealized "lab" situation for initial estimates only. I'll try to make the situation more realistic as I move forward.

So, let's assume the 4.2 changes are live.
DL - 8198 mana, 11733 Avg Base Heal, 1.15295 SP Coeff, K=1.32871, 2.5 sec cast time
HL - 2342 mana, 4400 Avg Base Heal, 0.4321 SP Coeff, K=1.32871, 2.5 sec cast time

So let's spam DL and assume the paladin has 8000 Spell Power.
Final Heal: 27845
HPM: 3.3967
HPS: 11138.1
MPS (Mana consumed per sec): 3279.08

Spam HL:
Heal: 10439.41
HPM: 4.457
HPS: 4175.763
MPS: 936.88

Consider DL Spam base case
Let's pretend that unmodified DL spam is "normal". So everything we're going to do, is to bring HL spam up to par with unmodified DL spam.

For comparative purposes 1 spirit = 0.62 IC mp5.

For straight throughput (i.e. what you would see in Int and Crit), it would take an additional 266.73% throughput to be equivalent to DL spam. So we can pretend that 2342.2mps or 11711 MP5 is equivalent to 266.73% throughput.

So what would this ridiculous amount of throughput require?
47835 Crit Rating (technically impossible because crit caps at 100% for sure, and IIRC Blizzard caps it at 75%)

38317 Spell Power (achievable with 29806 int)

34175 Haste Rating (but this would also require an additonal 786.9792 MPS or 3934.896 mp5 or 6347 spirit)

11711 mp5 is roughly equivalent to 18889 spirit.

31880 mastery rating but does not consider the deprecation factor. With deprecation, this should at least double.

So what the heck does this all mean? I'm really not quite sure. It's suggesting that in a substitution type model, spirit is the best stat? I need to think on this a bit more and I'll get back to everyone.

So I've been running some more Excel spreadsheets. I believe that the whole trying to consolidate all the stats into 1 single metric is the wrong approach. Instead, the whole

XXX to comfort
YYY to comfort

is the right way to go.

Stat weighting for healers isn't like stat weighting for DPS. Instead, to borrow from biology (my apologies if I screw up this metaphor), healers have limiting factors. I consider them to be:

1) Reflex fail. (too slow on LoH, cleanse, FoL, etc.)
2) Throughput fail. (Standard friendly takes more damage than healer can put out)
3) Longevity fail. (Going OOM)

So that means as healers, we need to know what's causing us to screw up and adjust accordingly. It should be said however, that all three of these limiting factors can and should be better adjusted for by proper spell selection. It's unlikely that 4.2 is going to release content where FoL is too slow, or sustained DL spam is insufficient.

What can be derived from HL/DL up/down grading is that at a certain points, stats are equivalent. Suppose we determine that a ratio of 50% HL and 50% DL is sufficient HPS to sustain the raid. At some ratio of haste, you can downgrade enough DL's to HL's that you'd have the same amount of "regen" as a certain amount of spirit. In this particular example, it's like 500 haste to 80 spirit (I don't have my spreadsheet in front of me, but I'll publish it once it's more clean). The point here is that haste is an awful regen stat, which we already knew. Likewise if we found we could sustain 1:1 HL:DL indefinitely, adding more spirit means we could cast more DL's thus giving spirit a "throughput" bonus. Spirit likewise is piss poor compared to haste.

What IS interesting is that spirit in crap compared to crit. Crit is actually better than spirit even fore regen! If a certain throughput is sustainable, then using crit to downgrade to HL beats straight spirit. The exact amount is dependent on the initial ratio of HL:DL. At about 80:20 crit's about the same as spirit. Anything with more DL, crit does better. The reason for this is because if a certain HPS is sufficient every DL crit is 8k mana. And crit scales in this regard better than spirit.

Crit is 2x as good as it used to be, bumping it up quite a bit. I don't see mastery getting much better since it's pretty harshly hampered by 50% elibility. I'd imagine the deprecation factor will change from 0.3 to 0.4, but we'll see once 4.2 goes live.

So with this in mind, I propose the following stat weighting:

INT->Spirit to comfort (~2k)->Haste to Comfort (~1050)->Crit->Haste->Spirit->Mastery

The specific numbers are for me personally. Your definition of "to comfort" will vary from mine, and I'll try to pin down some better threshholds as I do more study.
I don't see HL usage fundamentally changing. The way I personally heal is beacon on the tank, and throw appropriate heals around the tank while keeping an eye on the tank. If the tank is dipping low, throw a couple of direct heals or LoD a bunch of people. The HL change means I don't have to heal the tank directly as much anymore.

I heal 2 tank situations similarly. Spam 1 tank while keeping an eye on the beacon target and throw direct heals as needed. I'm not quite sure how the non-ToR hpally's are dealing with it (beacon swapping?)

There are a couple of tabs/sheets. The tabs of interest are config and general math. Character specific math is incomplete and there's some incomplete spirit calculations on the config tab.
Also, took a look @ t12 for us. Looks like a great set for holy paladins, but I haven't checked the offtier stuff yet. I also originally complained about the look (looks more magey than paladiny), but compared to everyone else, I'd say we're looking pretty good!

DK - Butterfly head
Druid - Michael Jackson hat
Hunter - Pirate Horns
Mage - Feather
Priest - Dog Cone
Paladin - Badass mage robe
Shaman - head sticking out of a flaming jaw
Warlock - Actually looks warlocky. Helmet's a bit funny.
Warrior - Not bad. Armored desert nomad!
Lylthe- thank you for maintaining this great thread; I've enjoyed reading it very much. I have two questions, however, and I hope you don't mind me asking them here. I have seen on several well geared builds (including your own) where the points are split between EJ and IJ. Would you mind explaining the thinking behind that? Also, I noticed you are gemming for socket bonuses instead of straight intel. Are you doing this for testing purposes, or have you made a decision that this is the way to go (at least for yourself) now? Thanks again for your time!
I have seen on several well geared builds (including your own) where the points are split between EJ and IJ. Would you mind explaining the thinking behind that?

I myself use the same build that Lylthe uses, and with my current level of spirit, I only need one point in EJ to ensure that my judgements will never miss. Therefore, it really comes down to the added range from IJ, as opposed to the self-heal from EJ. I value the extra range and the flexibility 1/2 IJ provides, although 2/2 EJ would be a very minor throughput increase, with the increased chance of being out of range of a judge.

I'm not sure how you'd be able to model this, to be completely honest. I'm going completely by my own experiences here, and using 1/2 in each, I've very rarely been out of range for a judgement when I want to use one. How much of this is a result of those last 5 yards, I'm not too sure.

My suggestion is to just try both builds out yourself. If you find yourself hampered by a 20 yard judgement, move up to 25.
Yeah pretty much what everyone else says about IJ/EJ.

Regarding gemming. I'm pretty sure I'm doing it wrong. Gemming for socket bonuses was something I thought was useful, back when I started doing the math a few months ago. My recent work suggests trading 10 int for at least 30 spirit, haste or crit. Do not trade for mastery.

If you're curious, I can get into the math a bit more closely, but a 10:30 tradeoff should be a pretty good guideline. I don't believe a 10:20 tradeoff is worth it. If you have additional questions, feel free to ask, since this is a somewhat esoteric subject when you get into the nitty gritty details.

This is speculative for 4.2, but given that crit at least is better in 4.2 than 4.1, I'm doing it wrong. I haven't gotten around to swapping my gems yet.
I was primarily keeping the second point in EJ to ensure 100% hit, but if I don't need it for that even at my spirit level, then I see no need for the second point just for the extra healing on my self (and the tank through Beacon?). Since I still scramble around in dungeons a fair amount, the increased judgement distance, albeit only 5 more yards, I feel will be very helpful, and since I see so many raiding pallys carrying at least one point in IJ, I assume it is needed in raids as well.
So, thanks for helping out another newbie and clearing that up for me :)

Join the Conversation

Return to Forum