Beacon judgement and SS take up 3-5 GCDs a minute, depending on if you have divine guardian and refresh jotp early, which I usually do. Lets make this a hypothetical fight where I holy shock twice a minute, and spend 2 other GCDs between hand of blank, DP, LoH, bubble yada yada yada. That’s 8 GCDs a minute, roughly 13% of my time. I suppose at that point a person could just multiply their current value for haste by .87 and call it a day, but they won’t be getting the full story.

The haste soft cap should be 676 rating (20.6%) which would put HL at a 1.333 second cast. I think 777 rating (23.7%) puts HL at the cap during bloodlust if anyone is interested, might not seem like a relevant cap to keep track of but we’ll spend a minimum of 6.66% of every fight under it.

Personally I don’t think I’ll go much beyond the cap as a 1.333 second HL is pretty damn fast, I’ll certainly try to hit it though. Going through Ulduar a time or two on not-hard-mode will give me a better idea of what to value for the hard modes, after that I can reevaluate and/or consult rawr (rawr always gives me a lower value to haste than I expect though). If fights turn out to be mostly calm with phases of absolute chaos I might end up singing a different tune.

Forgive me if this is slightly offtopic, but is it possible to parse a combat log (say through WoW Web Stats) allowing me to see every time my holy light overhealed by a certain percent? Like, I'd like to know whether my overheal has a tendency to be 20% for every cast that isn't 80%-100% overheal, or how many casts my overheal was 40% or higher, or something meaningful like that.

Essentially, if I overheal for an average of 50% for each of my casts that aren't 80%-100% overheals, then I wouldn't need so much spellpower and could drop some in place of int or haste, but if I overheal for an average of 20% for each of my casts that aren't 80%-100% overheals, then I might want to look into beefing up my spellpower. Make sense?

You could use the same analysis to figure out how much 5/5 Divinity would be worth to you, interestingly.

Saying that "if my spellpower almost always results in overhealing is useless" is a rather incomplete statement. It's not enough to show how little benefit 1 stat gives, you also need to show how much benefit you could've gotten if you replaced that stat with others. Just like extra spellpower can result in overhealing, more mana can result in more heals that overheal. You can't get any useful information by just evaluating overhealing for spellpower without evaluating overhealing for extra mana.

I disagree with you completely based on the fact that you used spellpower as your example. Spellpower is a static stat in every sense of the word, there is only one thing attributed to spellpower and that is the amount it heals for (via direct healing or SS absorbtion), comparing Spellpower to say Intellect is worthless because the value of intellect is dynamic in that it brings regen, spellpower, higher initial mana start point, crit. You cannot compare spellpower to a stat which has no value. The same can also be said of comparing spellpower to crit rating or haste rating, neither has a static value as they dip (into different areas) to give varying results.

Spellpower value is measured by taking the amount of effective healing above the base heal you have done in a fight divided by the amount you have (to give a value in terms of healing done). If you consider this against the statement you are arguing with I hope you see where it is in fact correct.
Haste can have both a positive and negative value but it is impossible to measure this either way. The only possible way to consider giving value to haste would be to replay the exact same fight over and over with different values of haste, each point of haste giving a different value in terms of healing done either positively or negatively. When we move to a different fight then we need to re-evaluate the value of each haste point.
Crit will always maintain a positive value (unless you never crit whereby the value is 0) and needs to be considered for both effective healing increased over a fight and the mana returns beneficial to the fight in question (if you'll never go oom crit mana regen has null value), the problem with valuing the mana return and extra healing comes in the same way that crit is worth to casters, each point is worth less than the last point in terms of overall benefit, 1 crit rating at say 10% crit is worth a lot more than 1 crit rating at 60% crit.
Intellect is perhaps the most difficult of the 4 stats to quanitify as it "double dips" into the mana regen stakes by both giving a higher return from replenishment but also providing "crit regen". It also provides a higher initial cast value (more int = more base casts till oom before regen mechanics) and more spellpower, intellect scales via the Divine Intellect talent and Blessing of Kings, the only of the 4 stats to do so. When you consider the value of intellect you have to at times discount the extra healing from extra crit and spellpower, extra regen from both mechanics and the higher initial mana pool when each is not required.

Another note when considering the value of stats would be that Spellpower is the only "leech" stat we have, the only interaction it has with our 4 other important stats is to take value from what they provide. The other 3 stats compliment each other in some way.

I'm going to close out this post there as it could possibly be considered rambling away from the initial reply which I intended to be "spellpower is a worthless stat when it stops giving decent returns".

Hard for me to respond to your entire post, but with crit it actually becomes better the more crit you have (along with your other stats). Essentially, the more mana you have to spend the more mana you get back from crit. Using your example, if we compare the gain of 1% crit when you have originally 10% crit vs 60% crit, assuming the same amount of available mana so lets say hypothetically 40000 mana to draw on. In your first mana cycle for the low crit chance, we're looking at 3000 mana back or 3300 with 11% crit, a gain of 300 mana. With the high chance to crit you're looking at 18000 or 18300 with 61% crit, again a gain of 300 mana. So at this stage in the game 1% crit gives the same benefit across the board. If you went through the second cycle and used the mana you got back from critting to keep casting, the extra 1% chance to crit gives MORE mana back when you already have a high amount of crit compared to when you have a low number.

If you wanna see more math, here's some off of the earlier example.


Crit Chance---------- 10%-----------11%---------60%----------61%
1st Cycle-------------3000-----------3300-------18000-------18300
2nd Cycle-------------225------------272.25-----8100--------8372.25
3rd Cycle-------------16.875--------22.461------3645-------3830.304
Total Mana Back-----3241.875-----3594.711---29745-----30502.554

Might look a bit messy, dashes had to be thrown in because I suck with forum formatting for the most part, but as you can see extra total mana back from 1% crit from 10-11% gave me 353 mana that I wouldnt have had otherwise. From 60-61% I get 758 extra mana that I wouldn't have had otherwise so the 1% extra crit is giving me nearly twice as much mana back in this scenario because I already have more crit. I'm assuming GSoW, 4pcT7 and libram of renewal for how much mana I'm getting back (rounding to 75%). If you have a different setup the principle remains the same, the more crit or just mana in general you have, the more effective that 1% extra crit will become.

Lets iterate on your results again, from 10% -> 11% crit the difference itself is 3241.875 mana to 3594.711 mana, the difference at 60% -> 61% is 29745 mana to 30.502.554.

Now lets consider these as a percentage "crit regen" increase. Rough napkin maths to follow:-

10% -> 11% crit change
3594.711-3241.875 = 352.836
(352.836/3241.875)*100 = 10.88% increase in "crit regen"

60% -> 61% crit change
30502.554-29745 = 757.554
(757.554/29745)*100 = 2.54% increase in "crit regen"

So although the numbers themselves are bigger, the relative value of crit has gone down by over 8% in terms of regen. To clarify, each percentage of crit we add allows for more "pure" mana regained obviously but drops relatively as you get more crit. It isn't an issue because we'll never have to worry about having 100% crit and still running oom, each percentage of crit we add will give us tangible returns on mana that we can see and therefore it is important to have more, but this doesn't mean that the value for each added point is greater than the last.

Yeah, you can have your statistics prove whatever point you want to make...

Do people actually read the first post?

There we find:


tl;dr version:
Crit scales with itself, therefor each point of crit is worth more the more you already have.

If you want to throw out "you haven't read the first post" comments then please make sure that you have actually read it through yourself please.

I agreed with this statement (by using it in my napkin maths) in my second post in which I worked out the percentage mana regen. The graph merely shows the amount of mana regained over 1 cycle "essentially", it does NOT show the value of crit, it is an indicator as to how much more mana 1% crit will give.

Will finish this when I get home 20 mins max

IE, you can't be bothered to read his post and respond intelligently, so you'll just copy/paste something that does nothing to directly refute his point.


To CrazyScot:

The reason your math doesn't work is because you're essentially comparing the mana regen increase to a greater "base" amount. In other words, lets say there's a basketball player who usually scores 10 points, but today he scores 11. Then there's one who averages 30 points, but today he scores 32. Sure, you can say the one who went from 10->11 increased his performance more, but you can't refute that the one who increased his performance from 30->32 had a greater marginal effect on the game, even if he didn't improve his base contribution by as much.

You contradicted the statement and are saying crit is worse the more you get, which is totally wrong. It doesn't show one cycle it shows all of them, it uses 1 / (1 - .6 * crit chance) to compute total mana reduction of crit.

What is the value of crit (as a regen stat) if not how much mana it gives you?

Moving the topic slightly away from the theoretical and more towards the practical: Assuming you're a Holy Light-spamming pally, how could you measure your overheal in terms that could actually tell you how important spell power is in relation to other stats?

Reiterating my previous post, if I'm overhealing more than 40% on each heal that lands without being 80-100% overhealing (these are subjective breakpoints), wouldn't that suggest I have too much spellpower and could be better served by haste/crit/mp5/int?

What I'd love to be able to do is take all my holy lights and graph them so that I can see where the biggest overhealing spikes lie (heavily near the 100% end, I'd imagine, with some other peaks around 10% and 40%). Overhealing as a total is fairly useless – according to my few minutes of research, a lot of top holy pallies run naxx with ~70-80% total overhealing, but that doesn't tell us where the overhealing comes from; is it preemptively healing a tank when he's got 100% health and mistiming it? Is it healing someone who's low on health only to have another healer top them up before your heal lands? Is it because you're using too much spellpower and the minority of your holy lights actually heal for their full amount?

I'm not sure if anyone's delved into this yet, but it seems like important questions our community needs answered.

TLDR: crit scaling with itself, as mentioned on the first page, applies to one model, which may or may not reflect your healing style. Understand what a model means, and remember that other people heal differently, and a different model may be better for them.

KYA1337, what the frost page doesn't state, is that that characteristic is only unique to crit under the following model of healing:

I presume the fight will last X seconds. I will attempt to maximize my raw healing output over this time.

This model has definite strengths, for one, you know when you're overspending and can rein that in. It's also completely unreliant on damage taken data, which is so incredibly volatile based on encounter, gearing, personal style and RNG that it's pretty much impossible to gather an adequate set of. It also means that if you're calculating it in any detail you might well have a strong impression of when you're overspending, and need to start using FoL more than normal to avoid going oom.

Strictly speaking to follow this model you should be constantly casting either FoL or HL, aiming for a fixed ratio of the two over the duration of the fight. No-one does that, so lets pretend FoL doesn't exist and maximize HL HPS. In that case, you end up with:

I have a fixed HLs per sec figure that I'm aiming for, if I stick to it I expect to be OOM after X (subject to RNG). I'm allowed to go overbudget temporarily, so long as i spread my HLs out more to bring things back in line.

This allows you to run a completely non-reactive (i'll term cancel-casting as semi-reactive, since reactive is a dirty word) healing strategy - I will press HL every 1.89 seconds. You can still adapt, a spike requiring you to chain 3 or 4 HLs results in you resuming at 1.95s per HL to make up the difference.

Now, how many of you who talk about effective mana actually do this? That's not rhetorical, I'm vaguely curious, it's what I suspect people mean when they claim to HL spam, because there's no way you can keep up a HL every 1.4s for 6 minutes (2-3 minutes on the other hand, which is closer to how long fights in naxx currently last, is doable). If that's what you do, it's a great model, stick to it, you're doing good.

I'm a cancel-caster though, my rate of expenditure of mana is related not to how much of it I have available, but to how much damage the people I'm responsible for healing have taken. So in that case, expenditure of mana can be approximated by a constant (determining this requires data, loads and loads of data, more than I have, so data mixed with guesswork, also there's averaging acroll different encounters in there, and ignoring of gimmick fights). So for me, effective mana is out, time to OOM is in. Suddenly, crit and mp5 scale with themselves in an identical way, int in an inferior way (still miles better than anything else obviously). For me, the crit scaling mentioned on the front page doesn't apply.

Edit: l2proofread

I was a little later in getting back than expected but here goes.

That is the intention of the maths, in working out the "real" value per point of crit you need to take in the effect it has on yourself, the value of one crit as a critical strike percentage is static (47 crit rating to 1%?), this does not mean that the value of crit rating is static, the value of one crit rating on your mana regen depends upon the current amount of crit you have. Which leads me to

You are correct in a sense that the value of crit in in how much mana it gives, but to assign a value on how much crit rating is worth compared to say 1 spellpower you have to take into account how much it effect it has relative to what you already have. Crit does get worse the more you get when you work it out relative value of it (as I showed with my napkin maths) BUT and this is the disclaimer I should have stated initially instead of assuming people would realise; this diminishing value will occur only when compared against our "strongest" or "most linear" stat.

And you are right, I was incorrect in saying that the graph was the same as the calculations shown earlier but the graphs look similar.

*Edit* As the post previous to this touched upon it, crit only scales upon itself for fights where time allows it to. Time is the one thing we rarely consider when we talk about crit when it is in fact the most important factor. For fights where mana is no issue crit is worthless as a regen stat, it is merely a throughput stat and not as strong as haste. The most optimal use of crit in the perfect world is to balance your crit/haste/spellpower to the point where you put out the most healing safely and finish the fight on 0 mana when the boss dies, the best way of doing this is to increase your haste if the fight is spiky or spellpower if the fight has a consistent level of damage (although each is useful in either situation).

Of course we can compare int to spellpower and to any other stat for that matter, that's the whole point of theorycrafting in the firstplace. If they're not compareable how would you ever choose items or even gems? Yes the comparison isn't 100% the same on every possible set of conditions, but using a standard set of conditions will give results that are more than good enough for most realistic conditions. At the end you only care about one thing - your ability to avoid raid members dying. Each stat contributes to that ability in one way or another, and that's what is getting compared here.

CrazyScot, your primary huge mistake is that you look at relative gains to a meaningless value. I don't care how much I increased my mana in relation to the mana I get back from crits. I don't care how much I increase my mana compared to my max mana value. The only thing I care about is how much I increased the amount of total mana I can spend during the fight compared to the amount of mana I could've spent over the whole fight before. That would be the sum of all non-crit mana factors (they're all additive), multiplied by crit mana return factor. I don't care how much mana I get from the extra crits compared to how much mana I was getting from crits, I care how much mana I get from crits compared to how much mana I already had total. Which one is more important and by how much depends on the fight.

Saying time is required for crit to do its job is also a pointless comment, as all theorycrafting assumes you're using all your mana. If you're not using all your mana, what's the point in more mana? We're maximizing ability to heal where our full healing power is needed, because if our full healing power is not needed we're already good enough for dealing with it. So efficiency considers how much you can heal using all your mana, and burst considers how much healing you can do in a short timeframe where mana doesn't matter.


While getting statistics for how much extra healing vs overhealing would be gained from additional spellpower, this data would be far from enough to compare it to other stats, especially mana stats. Keep in mind extra mana would result in you casting more HLs, which would then very likely result in a certain amount of overhealing as well. If you just say "only 10% of the spellpower doesn't overheal, so I'll give spellpower a 0.1X factor" you're making a big mistake, as "only X% of the mana you have doesn't overheal" is also true and you'll have to multiply mana effects by that factor as well. Just knowing how much of the spellpower is overhealing is a start, but it's far from enough to make stat comparisons.

Haste never has any negative effects (other than the loss of other stats that could've been gotten in its place). If you think slower casts would do more healing at any point, nothing stops you from casting slower. If increasing haste (without reducing other stats) reduced your healing it's purely your fault, as no matter how much haste you have you can, at worst, do the same healing as you did with no haste by casting the same number of spells at the same times. Haste, however, compared to other stats, has very little value when it comes to how much healing you can do by spending all your mana, while having a very high value when it comes to increasing your ability to do as much healing as possible over a timeframe short enough for mana not to matter during it. Plus haste acts as some sort of a slight reaction time buffer, though that is quite hard to quantify due to how situational (and probably small when compared to other stats or the other benefit of haste) this benefit is.

The value is anything but meaningless, it's just how you interpret the value yourself. I know for instance that should I keep the 51/0/20 build going into Ulduar that I shall be considering crit in the same way as I consider mp5, good, but not worth gearing for if I can get a better haste/int/SP upgrade as I personally value them more with my 55%+ Holy Light crit at raid buffs.

mp5 cannot be dwarfed by illumination, since it gets multiplied by it. That is, if I have X more mana per fight, that mana will be spent on casting more heals that will also return mana from illumination. What mp5 does get dwarfed by is max mana/replenishment/divine plea. Crit scales somewhat better than mp5 because it gives enough % per rating point that when it multiplies all of the mana sources (mp5, replenish, divine plea etc) the increase in total mana is bigger than when you add some mana mp5 (which is multiplied by illumination but is still not enough to win).

There's a very common misconception that stats scale differently depending on how much you actually cast. This is not true, since the number of casts is limited by the amount of mana you have, and if you're taking mana into consideration it means you're considering a fight where you have the possibility of needing all your mana, and therefore the number of casts is directly calculated by the amount of mana you have (and the time you can spend casting to decide on HL:FoL ratio, but aside from slightly different crit rates they both get multiplied by the same illumination so time you can spend casting doesn't matter much as long as mana matters). You can't say "on this fight I only used 1/2 my mana, so illumination wasn't as good compared to mp5 as it was on another fight". On that fight where you didn't spend all your mana, so you didn't do the max healing you can do with your current gear, and therefore the data from that fight is irrelevant for upgrading gear to maximize the healing you can do.

Number of casts is not dependant on the fight, even if you don't use your full healing abilities it doesn't change which stats are good for a fight where you do need all your healing ability, and it's your healing ability you're trying to maximize, you're not trying to maximize the mana you had leftover at the end of the fight.

"how much you actually cast" - What does this mean?
What does make a huge impact is "How much base mana cost spent per time unit"
mp5 is a static mana increase.
crit is a variable mana increase.
Short fights make int (max mana pool/fight time) and crit (constant HL spam) more powerful.

When people say "Illumination dwarfs mp5" what they are really saying is that under most fight conditions crit yields as much effective mp5 as the stat mp5, if not more, while yielding additional crit healing. Now if the fight is such that you are casting less then "1 HL per 2 seconds" then the stat mp5 will begin to yield more mana regen then crit. At that point you have to decide whether the additional healing from crits is preferred over the additional mana from mp5.

This leaves the dilemma, the one that plagues this entire thread, because there is no answer.
Assuming you are at the line where you have just enough mana to heal duration of the fight... improve your burst or improve your total healing done.?
Additional burst makes you more effective in a crucial moment, additional total healing done lets you spare heals elsewhere, or do more tank heal spam to make the tank damage smoother.

Just because WWS shows you a lot more mana returns from illumination compared to mp5 it doesn't mean anything about the value of crit as a stat compared to mp5.

The misconception is that people think that the rate at which you spend mana depends on complicated fight mechanic details / playstyle / whatever, but for gear comparisons what matters is the rate at which you CAN spend mana, not the rate at which you actually spend it. The rate at which you CAN spend mana is only dependant on how much mana you have available.

The only question that has no answer is "burst VS efficiency". From an efficiency perspective there's no argument about which stats are best unless you start making overhealing arguments which as I've explained are a lot more complicated than what people make them out to be and actually work against more than just the stat they claim is better/worse. When it comes to burst healing there's also no argument (other than the definition of "burst" - "reliable burst" or "average burst", though in reality both serve a purpose, depending on the fight).

The statements that claim crit scales differently depending on the fight have absolutely no basis. I mean yeah it depends on the fight but not in the way people claim it to be.

In short, please use rawr before you try to re-invent holy paladin theorycrafting....

I've seen (and shared) a lot of disappointment over current Ulduar itemization and the lack of crit/haste pieces and prevalence of mp5. i understand that we haven't seen all the available loot, etc, but just looking at what we have I'm starting to think that mp5 will be required because of the haste "soft" cap.

I added up the amount of haste values of all crit/haste Ulduar plate, mail and misc pieces:

Plate
Breastplate of the Devoted : level 239 Chest (XT-002-25-Hard)
Ancient Iron Heaume: level 226 Head (Iron Council-25)

Mail
Runeshaper's Gloves: level 226 Hands (Iron Council-25)
Belt of the Fallen Wyrm: level 226 Waist (Razorscale-25)
Bindings of Winter Gale: level 239 Wrist (Hodir-25-Hard)

Misc
Pyrelight Circle: level 226 Ring (Ignis-25)
Charm of Meticulous Timing: level 239 Neck (XT-002-25-Hard)

I then included T8 shoulders and pants (since the T7 4pc bonus is broken by the crit/haste "BIS" gear above) and filled in gaps with current 213/226 gear. Note that i did include two haste/mp5 items; the T8 pants and Charm of Meticulous Timing. The total amount of haste came out to 703 or 21.09%; well over the cap of 687 (20.6%). No haste gems are included.

Given that any haste beyond this will not lower the global cool-down below one second, why would we continue to gear pure int/crit/haste? It appears more reasonable to gear for crit primarily, haste secondarily until 687 is reached, and then mp5 after that. I wanted to check with you guys to make sure I'm understanding the haste cap correctly, as there are many knowledgeable players out there saying that all mp5 gear should be completely avoided and I'm having trouble wrapping my head around that given the above numbers.

No knowledgeable player would or has said that. Indeed, looking at the haste numbers we've already achieved it seems to make little sense to NOT take MP5 on gear.

Because when you mostly cast HL, you don't care much if haste doesn't lower your GCD further.