PvE Holy Paladin 4.0.3a Theorycrafting


6 Troll Shaman
Glad you found this useful!

Thanks for that. I'm mainly talking about live WoW testing I did a few weeks back regarding the SP coefficients and base heals. Anyways, thanks for the info. I'll try to find some time to do more testing.

Anyways, more on Crit vs. Mastery vs. Haste

Crit vs. Mastery vs. Haste

If you don't understand proper spell selection, don't even bother reading this. The benefit you gain from optimizing gear is NOTHING compared to good play. This pretty much applies to all classes and specs.

This discussion only applies to item budget. I'm not going to discuss raid buffed 40% crit vs. 100 haste rating. That's not fair because raid buffed crit includes crit from int and other random sources. This is for a gear to gear (reforge, gem, etc.) comparison. To be fair, we need to compare the <whatever> from rating ONLY to <whatever else> from rating only.

@lvl 80
45.91 Mastery Rating = 1 Mastery
45.91 Crit Rating = 1% Crit
32.79 Haste Rating = 1% Haste

100 Haste Rating = 3.05% throughput (measured in HPS) increase for haste
100 Mastery Rating = 2.18 % throughput (measured in HPS) increase for mastery
100 Crit Rating = 1.09 % throughput (measured in HPS) increase for crit. This is due to 2.18% increased chance to crit * 1.5 crit bonus.

Please remember, that these % increases are increases as if you had 0 of a certain stat. If you already have 900 haste, an additional 100 haste will boost 3.05% of what you would have with NO haste. Likewise, the same applies for discussions with crit and mastery. Crit and Mastery are more complicated because you already have some base crit from intellect and 3.3% that's just there, and you start with 8 Mastery just for being a holy paladin. The % increases don't multiplicatively stack with those base benefits.

Also, this assumes all bonuses are used to their full extent. Crits are not overheal. Mastery shields are fully absorbed. Haste is not GCD capped not is it overhealing.

So that's the easy part done.

Secondary (Good and Bad) Effects of Haste
The problems with haste come from, that while HPS go up, expended MPS (mana per second) go up accordingly. This can be counteracted by simply not casting. But this sort of wastes the haste. The unspoken benefit is that haste eases reaction time requirements. If someone's going to die right away, haste means the heal comes off quicker.

GCD capping cuts off benefits to haste since you can't cast any faster, but the reaction time benefit for casts still applies. GCD softcap for holy paladins in a raid environment is 894 haste rating.

Haste contributes to Overshield, which I'll discuss below.

Secondary (Good and Bad) Effects of Mastery
And this is the precisely the reason World of Logs is fail right now. Paladin Mastery usage isn't tracked right now.

Wasted shields, or what I call Overshield, occurs when:
1) A shield falls off in 6 seconds without being used.
2) Another heal lands on the target before the old shield is used. The weaker of the 2 shields is wasted.

Trying to predict/model overshield is REALLY hard to do, especially without logs to confirm the any models. Best I can tell, Case 1 almost always occurs when raid healing unless we have an aura fight.

Cast 2 almost always happens when healing the tank, because we can usually get multiple heals in between boss swings, and the tank is likely to avoid some attacks.

Exactly how much Case 1 and Case 2 apply, is anyone's guess, but if I had to guess, 2/3 or more of total shields are wasted.

What makes this MORE complicated is that you already have 8% shielding. For your mastery rating to be useful, that 8% must be eaten FIRST. So in a weak aura fight, if you had 15% shields, and the aura only eats 10%, then the 5%/7% of your shields are wasted.

Secondary (Good and Bad) Effects of Crit
Good Effects:
1) Keeps Conviction rolling
2) Procs Infusion of Light

Bad Effects:
1) Overheal

Overheal is the reason crit is bad. We can't pick our spells and HOPE for a crit. People die this way. So if a crit holy light could top someone off, but a normal wouldn't, we might as well use Divine Light. This is still a big deal in ICC. This might change in Cataclysm when it's ok for someone to sit at 50% health. If I had to guess. however, come Cataclysm, haste and spirit will still be better than crit, but we'll see.
Edited by Lylthe on 11/24/2010 3:33 PM PST
6 Troll Shaman
When does Divine Light overtake Flash of Light in HPS?

Divine Light @ 80
10652 Base Power
1.099 SP Coefficient

Flash of Light @ 80
6459 Base Power Avg
0.893 SP Coefficient

Fortunately, DL and FoL have the same % increases, so I'm going to bundle all the Walk in Light, Seal of Insight Glyph, etc into a single constant k. You can merge crit as an average increase into k as well. Again, crit effects DL and FoL the same.

I'm also going to bundle all haste effects into h. Take the individual effects (rating, bloodlust, JotP, etc.) and multiply them together to get h. 60% total haste means h = 0.6.

HFoL (How much FoL heals for) = k*(6459+0.893*SP)
HDL (How much DL heals for) = k*(10652+1.099*SP)

CFoL (FoL Cast Time) = 1.5/(1+h)
CDL (DL Cast Time) = 2.5/(1+h)

Before GCD Capping:
HPS FoL = k*(6459+0.893*SP) / [1.5/(1+h)]
HPS DL = k*(10652+1.099*SP) / [2.5/(1+h)]

Do the Algebra:
HPS FoL = (1+h)*k*(6459+0.893*SP) / 1.5 = (1+h)*k*(4306+0.595*SP)
HPS DL = (1+h)*k*(10652+1.099*SP) / 2.5 = (1+h)*k*(4260.8+0.4396*SP)

So before GCD capping, FoL will ALWAYS have more HPS than DL. But the GCD does cap FoL's cast time. So with that in mind

After GCD Capping (Note, you need 150% haste to GCD cap DL, not realistic):
HPS FoL = k*(6459+0.893*SP) / 1
HPS DL = k*(10652+1.099*SP) / [2.5/(1+h)]

k*(10652+1.099*SP) / [2.5/(1+h)] > k*(6459+0.893*SP) / 1

Do the Algebra
(4260.8+0.4396*SP) + h(4260.8+0.4396*SP)> 6459+0.893*SP
h > (2198.2 + 0.4534*SP)/(4260.8+0.4396*SP) Note (4260.8+0.4396*SP)>0 so no flipping

This is a function of Spell Power, so here's the appropriate table:
2000 SP : h > 0.6041
2500 SP : h > 0.6216
3000 SP : h > 0.6378
3500 SP : h > 0.6527
4000 SP : h > 0.6665

So the flip point should be anywhere between 60.4% to 66.7% total haste depending on your gear.

So how much haste rating and under what temporary haste buffs would you get 66.7% (worst case scenario) haste?

hr = haste rating

Under Bloodlust
1.6665 = (1.05 Raid Buff) * (1.09 JotP) * (1.3 Bloodlust) * (1+hr/3279)
hr = 394 Haste Rating

Under Divine Favor
1.6665 = (1.05 Raid Buff) * (1.09 JotP) * (1.2 Divine Favor) * (1+hr/3279)
hr = 700 Haste Rating.

EDIT: These numbers are lower with changes to speed of light in 4.0.3a, but I don't feel like recrunching the numbers with only 2 weeks left on Dreamwalker b4 cata hits.
If you have 394+ Haste, under bloodlust use DL when healing Dreamwalker.

If you have 700+ Haste, under Divine Favor use DL when healing Dreamwalker.

Otherwise use FoL.
Edited by Lylthe on 11/24/2010 3:17 PM PST
37 Blood Elf Warlock
@ Tiraslin, your spreadsheet uses the erronous Int information posted at EJ by a person who no longer has the time commitment to update the information. 1 Int = 15 mana.

Spellpower scaling is no longer the CastTime / 3.5 * 66/35 * 1.25 that I derived for usage at the start of 3.0. If you guys were using CastTime / 3.5 * 1.88, then that was totally wrong for 3.X. and even for pre 3.X when it was simply CastTime / 3.5. (Cast Time / 3.5 * 66/35 was for non-paladin healing spells, as the removal of Blessing of Light added a flat modifier to all paladin spells)

The math for Holy Shock in your sheet is off. Holy shock does not get 3.5% more healing per Crit, having never used Googledocs to make a sheet, I would assume it lies with the X+10% parts of your formula.

@ Lylthe
Haste, Crit and Mastery are linear, meaning that 3% haste will always be 3% more healing over time given that you do not run into cooldown or overhealing/overshield issues. You might be thinking of the old Illumination where you recieved greater then linear mana returns from Crit.

Your Coefficients are off from what I have mathed out. These are the numbers I calculated with no Gear, no buffs, no spec, and no glyphs:

FoL: ~86.3%
HL: ~43.1%
DL: ~115.3%
HS: ~26.87% (PTR and Beta Holy Shock, 33% less then Live Holy Shock)
WoG: ~20.9% per Holy Power

Your mana regen formula is also inccorrect, but I don't blame you, many versions of it are floating around, even from classes that have been using the formula for over a year now.

SpiritMp5ooc = 5 * (0.001 + Spirit * Sqrt(Intellect) * 0.003345) or:
SpiritMp5ooc = 0.005 + 0.016725 * Spirit * Sqrt(Intellect)
SpiritMp5ic = 0.0025 + 0.0083625 * Spirit * Sqrt(Intellect)

The difference is minimal, but the formula you used is outdated. The numbers on the paper doll are NOT accurate from a spirit based perspective. OOC and IC regen on the char sheet includes a base mana regen of 5% of Base mana. For paladins, base mana is 23422 at level 85, and 4394 at 80.

The Char sheet values are determined as such:
RegenMp5 OOC = 5% Base Mana + 100% of Spirit Regen
RegenMp5 In Combat = 5% of Base Mana + 50% of Spirit Regen

There are also some things comming down the pipe that will change your analysis, like the Speed of Light change, and some of the glyphs being reworked.
Edited by Faelyne on 11/20/2010 5:02 PM PST
85 Blood Elf Paladin
If you have 700+ Haste, under Divine Favor use DL when healing Dreamwalker.

Nice to see the math prove what I inferred through experimentation.
6 Troll Shaman
Thanks for the feedback. Constructive criticism of my work is always appreciated.

Regarding Crit.

Sorry if I wasn't clear. 3% crit means you'll have 1.5% increased throughput over time, relative to a character with 0% crit.

What I meant to say was "realistically" you will have say 15% crit from talents and intellect (just an example, but illustrative of my point). You will do 107.5% of your healing with 0% crit. 107.5% is now the baseline. Additional mastery and haste will scale off of 107.5% because they stack multiplicatively. So to be "fair" crit rating needs to scale off the same base for comparative purposes.

Now suppose you gem and reforge to add 3% crit to a total of 18% crit. You won't go up an additional 1.5% from 107.5%, but you will go up 1.5% from 100%. Your final throughput will be 109%. The % increase here is 1.4% whereas the linear increase is 1.5%. It's a subtle point, but for the sake of "fairness" I consider the multiplicative benefit since we're comparing to mastery and haste.

Regarding SP and Spirit
Guess it's back to target dummy testing. I'm pretty confident in my dummy tests, but I might have made mistakes. I'll post the actual test data when I get the chance to perform the tests.

Not looking forward to Spirit tests if the paper doll is wrong :(
37 Blood Elf Warlock
The paper doll is not wrong. The mana regen section just has more in it then just the regen from Spirit, basically just the 5% base mana as Mp5. If you mouse over the spirit stat, it tells you exactly how much it gives.

Regarding the Crit situation, that comparison is misleading. You are thinking of Increases your Crit% by a %, which it does not. Going from 107.5% healing to 109% is an additional 1.5%, or said another way, rating stats are additive with themselves.

It is much the same as haste. Lets say going from 10% haste to 20% haste. (1.5/1.1) / (1.5/1.2) = 1.09%.

If they instead worked at a % increase over current values, you would get exponential scaling.

When you take your analasys and apply it to spellpower, you see the distinction. If I have 1000 spellpower, and I get 1000 more spellpower, I get 100% more SP. Now If I had 2000 SP, I only get 50% more. The % increase only reveals the ratio between what you have and what you gained, not how much you gained. The slopes of all rating stats, and spellpower, are all linear.

1% haste is always 1% more healing over time. Crit/Mastery is the same way.
Edited by Faelyne on 11/21/2010 5:57 AM PST
6 Troll Shaman
Faelyne, you're absolutely right. I'm not disagreeing that 10% <whatever> is 10% (5% in the case of crit) increased throughput. What you're saying is completely correct. Everything stacks additively with itself.

I might be overcomplicating things, but I don't want people to mistakenly believe, that if they gem say, 10% haste, their throughput will increase 10% over what they ALREADY have, because it won't. It will additively boost the benefit of haste by 10%. Otherwise, like you're saying, you would see exponential scaling.

I apologize if I wasn't clear about that in my writing.

Thanks for the info about spirit. Pretty subtle, but I'll get that updated. And yeah, I've heard about some of the incoming changes, and they will change things up.
6 Troll Shaman
Since there's some contention regarding SP coefficients, I'm posting in-game test data. The analysis on this data isn't 100% accurate, because I'd need more in-game test data. I'm going to base my analysis off some Wowhead numbers, which may or may not be accurate.

Tests were performed using these specs and glyphs:

I should have removed my glyphs, but I didn't realize it until too late. Seal of Insight was active.
At the time I had
INT - 2345
SP - 3428

My gear was my full holy gear with the exception of the Ring of Ashen Verdict, since that procs SP. I stood in front of a target dummy and healed myself. The results are posted here. I'm posting non-crit numbers only. I'm also considering averages only. 80-100 data points should give an accurate average, but I don't trust the absolute ranges

100 - Non Crits
17492 - Avg

100 - Non Crits
11499 - Avg

100 - Non Crits
6540 - Avg

82 - Non Crits
6127 - Avg

102 - Non Crits
3328 - Avg

I ran the SP Coefficients against Wowhead's bases, and I'm not getting numbers that make sense. My back derived numbers are within 5% of Faelyne's, and using Wowhead's #'s I'm getting numbers that are off by 20% or more. When I get the chance, I'll rerun this test at much lower SP values then solve the 2-var system.
100 Human Paladin
It occurs to me that a good way to establish base healing would be to remove all talent points, and gear, and cast naked. Well, in some nice non-stat rp gear, for modesty. I might try this if I feel ambitious.
85 Dwarf Paladin
Is it possible to see your screenshot Lylthe? I just got back to wow and I tried to see how do you place button. Mods would be helpful as well! Thanks for great guide.
6 Troll Shaman
Remember that you still have some intellect, and hence Spell Power. So we still have 2 unknowns, the base avg. and the SP coefficient you need to solve for (hence the need for at least 2 sets of data).

Since we want numbers for Holy Shock (Light of Dawn too) we need to take some talents. So we need to account for Walk in Light, and ugh, Conviction (keeping it up 100% or 0% of the time for testing sucks). It's not impossible to do, but it's easy to screw up. Including these coefficients into the math isn't hard provided that we're consistent with them.

The other problem with being naked is a practical one. It really sucks to have to drink after every cast :P

If you do end up doing this test, please post the numbers! More data points make me happy.

I don't have access to WoW right now, so I can't give you my exact setup, but here's what my UI looked like before 4.0.1 hit

It's by and large the same, but my buttons are different.

I try not to talk about addons too much because everyone's setup is very different. Your setup probably won't work for me, and mine probably won't work for you.

However, if you intend to raid, you need a minimum of 2 addons, a healer addon and a raid mechanic addon. A third threat meter is highly recommended, but not as critical.

Any Healer Addon - This means Healbot, Grid + Clique (this is a 2 addon combo), or Vuhdo. There are other options, but these are the 3 most common.

Any Raid Mechanic Addon - This means Deadly Boss Mods, BigWigs or any other equivalent addon.

Here's my personal list of addons:
Deadly Boss Mods
Power Auras (track procs/personal timers)
IceHUD (track holy power. I turned off everything else in IceHUD)
Recount (don't look at this during a fight. It's a debugging tool for wipes.)
GearScore (because even if people don't like it, PuG's still use it)

Non raiding addons are:
Addon Control Panel (let's me turn addons on and off w/o relogging. Still need to reload screen)
Edited by Lylthe on 11/22/2010 10:44 AM PST
6 Troll Shaman
EDIT: If you take this section and interpret it such that spirit goes down on the stat priority list, I'm going to smack you.

Why I love/hate Spirit
I personally hate spirit. Spirit doesn't regen nearly enough, but it's important because we have no other options. Yay for being stuck between a rock and a hard place.

A ToR rotation that uses 1 HS, 1 WoG, and 2 DL will burn ~ 500 mana per SECOND. Weaving in FoL will increase that number, and weaving in HL will decrease that number. At that rate of consumption, with 40k mana, and no sources of regen, you will OOM in 1 minute, 20 seconds. (LK has a 15 minute enrage timer. Net consumption of 44.4 mps to avoid OOMing.)

Non-spirit regen:
~200 mpFIVE from Replenishment in ICC gear
~35 mp5 from Seal of Insight, if allowed to melee
~75 mp5 from Insightful Earthsiege Diamond
~220 mp5 from base regen
~92 mp5 from Blessing of Might

~622 passive mp5 =~124.4 mana per SECOND

So with this in mind, with 40k mana, you will OOM in 1 minute, 46 seconds. And this is why paladins have mana issues.

How much does spirit help? with ICC level intellect, provides ~0.4 in combat mp5 per point in spirit on the low end. A best case scenario, we'll let spirit give 0.5 mp5/spirit.

To last 15 minutes on a LK fight w/ 40k mana, you would need 332 mpSECOND, or 1660 mp5 from spirit to sustain the ToR rotation I described. An optimistic approach would mean you would need 3320 spirit to get this level of regen.

You're not going to get this much spirit, even if you stack the wazoo out of it.

So... what can we do about this? Replacing HL for DL in the same ToR rotation means you consume ~191 MPS instead of 500. And to sustain this level of consumption over 15 minutes, means you only need 111 mpFIVE from spirit. This is much more reasonable and quite doable.

Would Divine Plea save your ass? When used on CD and if glyphed, it will give you an additional 50 mp5, or 10 mps. You're also eating the mortal strike effect. So does it help? Definitely. Is it enough to keep you from OOMing? Not even close.

And this is the exact reason why I say your spell selection is MUCH MUCH MUCH more important than stat weighting. Doing nothing but HL will give you a dead tank. Doing nothing but DL and FoL will OOM you in under 5 minutes. Being able to simply NOT CAST is the best way to save mana, and why I say you will violate ToR cycles.

Spirit helps, but it's not enough. Your playstyle/spell selection will be the reason you avoid going OOM, not your gear.
Edited by Lylthe on 11/22/2010 5:02 PM PST
37 Blood Elf Warlock
Judging SoI is ~275 mp5 if used on cooldown, in addition to any SoI procs, don't forget about that as it is one of our largest sources of mana if needed.
6 Troll Shaman
I wasn't aware that judging SoI for mana was live already.
Edited by Lylthe on 11/22/2010 5:00 PM PST
37 Blood Elf Warlock
Hmm, yea you got me there. I was just assuming it was, since I am playing alts these days, just off the top of my head I recall my mana going up when judging it, but it might have just been the judgment procing the SoI, instead of the scripted mana return :(
6 Troll Shaman
So 4.0.3a has hit. I haven't tested anything at all yet, but here's what I believe the patch will change for holy paladins. None of this has been tested yet, and is all conjecture based on the patch notes here:

Basically nothing has changed.

We're probably going to replace the glyph of beacon of light with glyph of light of dawn, but I don't see any fundamental changes.

Light of Dawn will HOPEFULLY be mildly more useful, but I'd have to see the numbers first. It certainly will make us a bit more hesitant to use Word of Glory, but more options is a good thing.

Protector of the Innocent and Blesssed Life are PvP changes and don't affect PvE holy paladins.

The changes to mastery, while interesting, probably aren't enough to warrant reforging for it. For test purposes, I will be going into HM ICC10 stacking mastery. I'll let everyone know the results when I have the data.

I don't see any notes about changes to Speed of Light, Infusion of Light, and Seal of Insight, the 3 changes I expected that would have significant impact on holy paladins.


Thanks for all the support. This thread is now officially "Highly Rated"! :)
85 Dwarf Paladin
Thank you Lylthe for the sreenshot and mods advice, it's really great to help me back a bit on what I've lost :).
6 Troll Shaman
Bah, my guild made me tank tonight, so I don't have the mastery test data yet.

The new Speed of Light has changed our haste soft cap to 894 instead of 1019. Also, the ToR rotation is simplified since we don't have to deal w/ Speed of Light as a proc.

So here are the changes that I'm seeing that aren't in the patch notes:
- Holy Shock heals got nerfed a bit. It doesn't look like it changes our rotation though. This makes me a sad panda.
- Speed of Light changed to flat 3%. This looks like a buff, especially with ZERO haste on T11 :(
- Glyph of Lay on Hands got moved to Major Glyph from minor glyph.
Edited by Lylthe on 11/23/2010 10:22 PM PST
6 Troll Shaman
In the event I don't have the opportunity to evaluate mastery this week, here was my plan if anyone else wants to take a shot at it:

Purposefully reforge all my crit/excess haste (beyond 894) into mastery. Since World of Logs doesn't track mastery, I'll need to use Recount.

Take Total Effective Healing + Overhealing - Illuminated Healing. = Total Healing

Total Shields = Mastery % * Total Healing

Compare Effective Illuminated Healing with Total Shields.


This isn't enough though, since 10% of my Total Shields is inherent. Determining the Shielding due to Rating means eliminating the inherent shielding.

There's no good model for that, but there are 2 extremes.

The first extreme, is the worst case scenario for Mastery Rating. The only rating shields occurred after inherent shield were eaten.

So Shielding Due to Rating = Total Effective Shield - Total Healing * 0.1 (Total Inherent Shields)

The best case scenario for mastery rating is that effective shields either were completely used, or completely missed.
So Shielding Due to Rating = Total Effective Shield * (% Master Due to Rating / % Total Mastery)

That gives you a range for total effective shielding. Now we can take a stab at comparing to crit/haste for throughput, and "saved" healing for spirit.
This topic is locked.

Please report any Code of Conduct violations, including:

Threats of violence. We take these seriously and will alert the proper authorities.

Posts containing personal information about other players. This includes physical addresses, e-mail addresses, phone numbers, and inappropriate photos and/or videos.

Harassing or discriminatory language. This will not be tolerated.

Forums Code of Conduct

Report Post # written by

Explain (256 characters max)