WotLK Healing Compendium v3.0 [theorycraft, specs, etc] - Page 12

Liths · 11/12/08, 9:18 AM

Originally Posted by Woozle

To be honest, I'm not sure exactly sure what hard data you're looking for that would convince you either way. SOL is tied to your crit rate. That means that its effectiveness is tied not only to the particulars of your spec, but also to the type of fight you are in and the types of spells you cast. Right now, I almost always get a SOL proc whenever I cast COH, whereas when I'm chain casting gheal, its a less frequent occurrence. Further, the number of talent points you have at 80 makes it very easy to pick up both lightwell and SOL without sacrificing too much. The tradeoff is more throughput on your base spells, for situational tools that increase your mana efficiency and/or allow you to do multiple things at once. Even if you could model it, it seems to me to come down to a personal choice and playstyle issue.

It kind of bothers me that people are saying that you get a SoL proc on almost every CoH, it's simply not true. Very easy to show with some simple math.
20% crit rate at level 70: 0.9^5 = 59% chance to not get a SoL proc.
Even at 25% crit with the glyph at 80 it's hardly going to be on every cast: 0.875^6 = 45% chance to not get a sol proc. The only reason you say that is that when you use it you normaly spam it a couple of times, once you cast it four or more times in rapid succession do you start reaching the almost every time stage.

Woozle · 11/12/08, 9:50 AM

Originally Posted by Liths

It kind of bothers me that people are saying that you get a SoL proc on almost every CoH, it's simply not true. Very easy to show with some simple math.
20% crit rate at level 70: 0.9^5 = 59% chance to not get a SoL proc.
Even at 25% crit with the glyph at 80 it's hardly going to be on every cast: 0.875^6 = 45% chance to not get a sol proc. The only reason you say that is that when you use it you normaly spam it a couple of times, once you cast it four or more times in rapid succession do you start reaching the almost every time stage.

Anecdotal evidence is not meant to be taken as actual evidence, and I'm sorry you mistook it as such. I was illustrating the relative frequency of occurrence of events between two different cast styles, COH vs. geal spam. My intent wasn't to provide cold hard data, it was to show that modeling SOL vs. lightwell is a near impossible and situation-variant task that makes it not worth the time. Further, I didn't think such a model would actually convince anyone of anything, since the mechanics modify playstyle, not throughput and thus don't respond well to modeling the same as spirit / int conversions (for example). Further, I view the endless arguments about SOL and lightwell as (nearly) pointless, because the choice between the two isn't the only choice out there, you can sacrifice throughput spells instead, so taking the narrow view of SOL vs. lightwell seemed to be choking this thread, and perhaps my response was a passive aggressive response to try and end it.

edit: the argument between SOL and lightwell isn't pointless, I just feel that at this point new posts in the same vein aren't adding anything to the discussion, this one included.

Llnith · 11/12/08, 3:02 PM

Have you done any research into the impact of the new regen mechanics combine with the [Darkmoon Card: Blue Dragon], or [Bangle of Endless Blessings]? I heard roomers they scaled down the bangle as your level increases but I haven't heard anything about the blue dragon.

Merple · 11/12/08, 3:05 PM

Originally Posted by Llnith

Have you done any research into the impact of the new regen mechanics combine with the [Darkmoon Card: Blue Dragon], or [Bangle of Endless Blessings]? I heard roomers they scaled down the bangle as your level increases but I haven't heard anything about the blue dragon.

Blue Dragon is and always has been 2%. It's stated so on the item. The Bangle on the other hand has a PL formula in it, indicating that 70 is 15% and loses 1% every 2 level until 0% at level 100.

As such, you'll still see some benefit to levelling with the bangle (great for spirit classes) and some benefit to using it at 80, but the Darkmoon Card will always be just as good as it was at 60.

Liths · 11/12/08, 3:17 PM

Originally Posted by Woozle

Anecdotal evidence is not meant to be taken as actual evidence, and I'm sorry you mistook it as such. I was illustrating the relative frequency of occurrence of events between two different cast styles, COH vs. geal spam. My intent wasn't to provide cold hard data, it was to show that modeling SOL vs. lightwell is a near impossible and situation-variant task that makes it not worth the time. Further, I didn't think such a model would actually convince anyone of anything, since the mechanics modify playstyle, not throughput and thus don't respond well to modeling the same as spirit / int conversions (for example). Further, I view the endless arguments about SOL and lightwell as (nearly) pointless, because the choice between the two isn't the only choice out there, you can sacrifice throughput spells instead, so taking the narrow view of SOL vs. lightwell seemed to be choking this thread, and perhaps my response was a passive aggressive response to try and end it.

edit: the argument between SOL and lightwell isn't pointless, I just feel that at this point new posts in the same vein aren't adding anything to the discussion, this one included.

I'm well aware of that, and it wasn't really directed at you and your post only. My point is, you won't be getting a SoL proc every time you cast a CoH. I mainly posted the numbers to show that SoLs value is going to become diminished with CoH on a 6 sec cooldown.

Isin · 11/12/08, 3:24 PM

Originally Posted by Merple

Blue Dragon is and always has been 2%. It's stated so on the item.
...
the Darkmoon Card will always be just as good as it was at 60.

Your first statement is correct, but the second is not really true. In WOTLK, the coefficient on int/spirit regen will drop dramatically as we level, and consequently we will get much less spirit based regen at 80. From a tester thread on the beta:

WotLK Beta (US-English) Forums -> Please do something about priest mana regen..

The coefficient on spirit regen will drop to about 60% of its effectiveness at 80 compared to what it was at 70. Priests will be much more reliant on Max mana based regen mechanics, such as shadowfiend, mana tide totem, and replenishment than on their personal int/spirit based regen in comparison to before. The Darkmoon card will still be 2% of your spirit based regen, but that overall spirit based regen will actually be lower at 80 than at 70 given comparable stats, and you are probably much better off going with a different type of trinket.

This is one of the reasons that const and dwarfpriest and all are weighing spirit so low in comparison to other stats. It is an ok stat for holy priests because of its synergy with other talents, but it is weak now in terms of its contribution to mana regen.

Crosshairs · 11/12/08, 5:41 PM

Originally Posted by Llnith

Have you done any research into the impact of the new regen mechanics combine with the [Darkmoon Card: Blue Dragon], or [Bangle of Endless Blessings]? I heard roomers they scaled down the bangle as your level increases but I haven't heard anything about the blue dragon.

I started looking at the [Darkmoon Card: Greatness] as an alternative. However I am not sure spirit will be my top stat so proccing 300 spirit and then trying to cheat the 5sr, might not be possible.

Starfire · 11/12/08, 6:41 PM

Originally Posted by Crosshairs

I started looking at the [Darkmoon Card: Greatness] as an alternative. However I am not sure spirit will be my top stat so proccing 300 spirit and then trying to cheat the 5sr, might not be possible.

You can proc 300 intellect though. Interestingly enough, its theoretically possible for this trinket to be 90 spirit / 300 intellect proc or 90 intellect / 300 spirit proc.

Seems like it'd be a good pick, with the Spirit proc being viable for Holy. But to your point, you can wait for a 300 intellect proc and then hit your shadowfiend.

Sinndir · 11/12/08, 6:47 PM

Originally Posted by Crosshairs

I started looking at the [Darkmoon Card: Greatness] as an alternative. However I am not sure spirit will be my top stat so proccing 300 spirit and then trying to cheat the 5sr, might not be possible.

Crosshairs, do you know if the +90 agility is there for sure or is it +90 to a stat of your choosing then procs your highest stat? One way or another as a priest it should proc either intellect or spirit both of which increase our regen.

Originally Posted by Isin

This is one of the reasons that const and dwarfpriest and all are weighing spirit so low in comparison to other stats. It is an ok stat for holy priests because of its synergy with other talents, but it is weak now in terms of its contribution to mana regen.

It is good that they are weighing spirit low. A Disc priest sees very little use from spirit, as they should have a high if not 100% I5SR casting time. They gain a bonus 5% from enlightenment and only gain spirit form regen from it. As a Holy priest there are more ways to cheat the rule (via free spells) as well as a direct conversion to spell power. However, for both Holy and Discipline, intellect provides us with more opportunities to gain benefits from each point. The top end mana increase, as stated above, increases our shadow fiend, replenishment, and mana tide totem (as well as starting with 20k mana will be super!). Not only that but with a direct conversion to crit strike and also increasing our passive regen it seems to just be better in a 1 vs. 1 situation. Discipline priests also must love the +15% bonus from mental strength.

Originally Posted by Liths

It kind of bothers me that people are saying that you get a SoL proc on almost every CoH, it's simply not true. Very easy to show with some simple math.
20% crit rate at level 70: 0.9^5 = 59% chance to not get a SoL proc.
Even at 25% crit with the glyph at 80 it's hardly going to be on every cast: 0.875^6 = 45% chance to not get a sol proc. The only reason you say that is that when you use it you normaly spam it a couple of times, once you cast it four or more times in rapid succession do you start reaching the almost every time stage.

Liths, this is the last thing I want to comment on. I think your math is a bit off. You would need to use a probability equation (Binomial Distribution, I believe) to determine the chances that you would have to get a crit first, then check that chance vs. Surge of Light.

If you had a 20% crit chance with a glyphed CoH (6 targets) you would be looking at six chances to get a crit (SoL proc), each individual in themselves.

I'm still having a tough time figuring out the correct model to use. (Havoc any suggestions?)

Dagma · 11/12/08, 7:05 PM

Originally Posted by Sinndir

Liths, this is the last thing I want to comment on. I think your math is a bit off. You would need to use a probability equation (Binomial Distribution, I believe) to determine the chances that you would have to get a crit first, then check that chance vs. Surge of Light.

If you had a 20% crit chance with a glyphed CoH (6 targets) you would be looking at six chances to get a crit (SoL proc), each individual in themselves.

The original post may be correct. The implication is that there is a half chance on each crit, the chance of each NON-proc is (1 - C/2), where C is the crit chance.

Chance of a proc, on a single hit: $\Pr(1|C) = C/2.$

Chance of a single non-proc: $\Pr(0|C) = 1 - \Pr(1|C) = 1 - C/2.$

Chance of n non-procs: $\Pr(0|C,n) = \Pr(0|C)^n = (1 - C/2)^n.$

And that looks the same as the implied calculations above.

Sinndir · 11/12/08, 7:26 PM

Originally Posted by Dagma

The original post may be correct. The implication is that there is a half chance on each crit, the chance of each NON-proc is (1 - C/2), where C is the crit chance.

Chance of a proc, on a single hit: $\Pr(1|C) = C/2.$

Chance of a single non-proc: $\Pr(0|C) = 1 - \Pr(1|C) = 1 - C/2.$

Chance of n non-procs: $\Pr(0|C,n) = \Pr(0|C)^n = (1 - C/2)^n.$

And that looks the same as the implied calculations above.

Hmm that looks very good. Perhaps Liths looked incorrect because of the reasoning. I personally am trying to get a model of the chance per cast of CoH that would result in a SoL proc. I guess you could just take the opposite of your answer. Something like this perhaps?:

C = Crit Percentage (where 1.0 = 100%)
n = number of chances to proc (n = 2 for Binding Heal, n = 5 for CoH, n = 6 for glyphed COH)

Chance of n procs: $\Pr(0|C,n) = \Pr(0|C)^n = 1 - (1 - C/2)^n.$

Example #1:
Binding Heal
C = 21%
n = 2
$\Pr(0|C,n) = \Pr(0|C)^n = 1 - (1 - 0.21/2)^2.$
$\Pr(0|C,n) = 1 - (.801)$
$\Pr(0|C,n) = 0.199$
Or a 19.9% chance to get a SoL proc.

Example #2:
Circle of Healing
C = 21%
n = 5
$\Pr(0|C,n) = \Pr(0|C)^n = 1 - (1 - 0.21/2)^5.$
$\Pr(0|C,n) = 1 - (.5742)$
$\Pr(0|C,n) = 0.426$
Or a 42.6% chance to get a SoL proc.

Example #3:
Circle of Healing glyphed
C = 21%
n = 6
$\Pr(0|C,n) = \Pr(0|C)^n = 1 - (1 - 0.21/2)^6.$
$\Pr(0|C,n) = 1 - (.514)$
$\Pr(0|C,n) = 0.486$
Or a 48.6% chance to get a SoL proc.

This seems right as between examples #2 and #3 the chance increases which is right because as you get more chances to have the outcome you want (a SoL proc) the higher the chance of getting one should be.

Imua · 11/12/08, 7:40 PM

Originally Posted by Dagma

The original post may be correct. The implication is that there is a half chance on each crit, the chance of each NON-proc is (1 - C/2), where C is the crit chance.

Chance of a proc, on a single hit: $\Pr(1|C) = C/2.$

Chance of a single non-proc: $\Pr(0|C) = 1 - \Pr(1|C) = 1 - C/2.$

Chance of n non-procs: $\Pr(0|C,n) = \Pr(0|C)^n = (1 - C/2)^n.$

And that looks the same as the implied calculations above.

That's it exactly. The math itself is trivial as well. (Well, trivial to put into a computer)

Is there a good way to put a table in here?
Just for reference:

First % is non-Glyphed. Last column on the right is if you have Glyphed CoH:

Crit%    % of Proc    % of Proc
5.0         11.89         14.09
10.0        22.62         26.49
15.0        32.28         37.36
17.5        36.73         42.27
20.0        40.95         46.86
22.5        44.94         51.13
25.0        48.71         55.12
30.0        55.63         62.29
35.0        61.78         68.47

So yeah, at 20% Crit rate on a Glyphed CoH, you have a 46.86% chance to proc SoL. Again, if you spam it twice in a row, you're not guaranteed 100% to proc, but like almost 75%.

(Just for fun, you hit 50.0% chance to proc SoL at 25.89% crit chance on a non-glyphed and 21.82% crit chance on glyphed)

Havoc12 · 11/12/08, 8:31 PM

People are forgetting a critical factor with SoL. Overwriting procs. Although the calculation takes overwriting per cast of CoH into account it is very difficult to make a reasonable estimate of how many overwrites you get from multiple casts of CoH.

What I can assure you is that 1 SoL use per 5 seconds (you DONT care about how many procs you get) is the absoludely limit of usability. Its in fact absurdly low. The most reasonable estimate of SoL is 1 use per 10 seconds.

Strings where you get multiple procs in a couple of casts then no procs for 3 or 4 casts are fairly common and they result in very high proc wasting

constantius · 11/12/08, 8:58 PM

You do realize that 1 use every 5 seconds is effectively 1/3 of your casts being free? It doesn't matter. If I get a free Flash Heal every 2 CoH, that's entirely sufficient. It's more than sufficient. I'm unlikely to use more than that.

Razzberry · 11/13/08, 11:30 AM

It seems like it would be useful to discuss [Ember Skyflare Diamond] and [Revitalizing Skyflare Diamond] in the meta gem section as well.