Efejel

Are you adding latency in to that calculation? If not, the simplifying assumption of averaging the cast times is probably what's giving the variation between your results & mine.

I use a calculation of the expected mean DPS contribution.
We'll say "p" is the probability to crit, and "q" is the probability to not crit.
d = average damage of a starfire hit.

(((d/3)*pp)+((2d/3)*pq)+((d/2.5)*qp)+((2d/2.5)*qq)) * .99 (chance to hit) = expected DPS

Of course the latency terms are added to the casting times, so 3 -> 3.2~3.4 and 2.5 -> 2.7~2.9.

I suppose with the Mystical Skyfire active I could add an addt'l 2 terms, r = proc, s = no proc and get
((((d/3)*pp)+((2d/3)*pq)+((d/2.5)*qp)+((2d/2.5)*qq))s + (((d/1.5)*pp)+((2d/1.5)*pq)+((d/1.5)*qp)+((2d/1.25)*qq))r) * .99 = expected DPS

 

doogless

A lot of this discussion of Moonkin DPS stemmed from the question of whether or not it's worthwhile to bring a Moonkin to a raid. By dropping Improved Faerie Fire, you've eliminated one of the strongest reasons for a Moonkin to be in a raid group rather than another ranged DPS class.

The balance and resto trees are bloated, but that's for another thread.

 

Efejel

3% hit = how much DPS? As has been pointed out, it only works if your melee DPS gears around the assumption that iFF will ALWAYS be up on the mob. In the event you're using a Feral tank it will also, I believe, prevent them from applying FF(F) with the message "A more powerful spell is already active" and thus take away rage-free threat gen. (This also assumes your tank is rage-starved, which is next-to-never the case for a Feral tank, except for possibly during the early going on Gruul if the Druid is OT and not MT). It also eats into a Moonkin's cast time and mana pool, albeit slightly.

The upshot: iFF is an argument for bringing Moonkin on a raid if their DPS is subpar. Personally, I'd rather demonstrate that in the upper stratosphere of gear a Moonkin has roughly the same DPS as a Mage on their own, and let the 5% crit aura do the arguing.





I found an error on my spreadsheet that was doing Spirit * 1.1 instead of Spirit * 0.1 in calculating the total spell damage. Face red, apologies, etc. Fully buffed my number which was 1538 ---> 1198 (before WoA).

The new Starfire DPS number I get is 1340.8 DPS chain casting, 1399.8 modelling meta-gem procs.

With 0.2s latency: 1300.1
With 0.4s latency: 1214.2
Both including modelling meta-gem procs. For safety's sake I rounded (2.5/2) = 1.5, though I'm not sure if 1.65 is more accurate than 1.9 with latency--I don't have the meta gem to do any testing.

 

doogless

3% fewer misses on shield slam/sunder/revenge/etc is not trivial in terms of threat generation.

I'll agree with you on this, though. If moonkin can reach the same DPS as other ranged casters it makes the argument for a raid spot a lot more convincing.

Anyways, sorry for the derail, resume the moonfire spam/theorycrafting

 

Efejel

Spellfire + Antler's of Malorne (Myst. Sky. Diam.) = 1460.9 DPS (0 latency) 1356.7 DPS (0.2) 1266.9 (0.4)
Spellfire + Spellstrike (ignoring 2 pc. Ss procs) = 1427.2 DPS (0 latency) 1331.9 DPS (0.2) 1248.6 (0.4)

 

Caliane

Assuming cos is up. Do you out CoS up on everything? hydross adds?
And it still takes a large amount of +damage for starfire to catch up to wrath even with cos.

 

Caliane

You are factoring chance to hit after crit. This is not accurate. Doing so reduces the value of crit, as it assumes a 10% chance to miss, would then reduce damage from crits by 10%.
Crit % however is seperate.
A spell can have 3 outcomes, hit, miss, or crit.
A simplified example would be: with 50% crit rate, 10% chance to resist, and 10000 casts.
The average result would be:
5000crits,
1000 resists
4000 hits.

Not,
1000resists
4500 crits
4500 hits.

Play a duelwield class vs level 73 bosses. 40% glancing blows, 29% miss pre+hit, 6%dodge/parry all create a "crit cap", but your crit rate is not reduced by 75% due to these factors. Your crit rate stays the 30-40% of the total attacks made, it is simply "capped" if you have a higher crit rate then the cap allows, thus the need for +to hit to increase your crit rate for DW classes past 25%.

 

Efejel

For melee tables, hit/miss IS calculated after crit.
For caster tables, this is NOT the case.

http://www.worldofwarcraft.com/info/basics/spells.html

Note how they never factor Eyonix's crit rate into the chance to hit calculation.

 

Caliane

Absence of mention of crit into the statement of +hit in no way asserts any relation of how crit is factored.

Hit/miss is not calculated after crit for melee either. Crits do not override misses for them either.
They are each factored seperately. In the above statment, you will crit 50% of the time, resist 10% of time, and hit 40% of the time.

Your crit will be "% of your total casts", not "% of total casts that are not resisted".

 

Efejel

Re-write the equation with the hit term where you think it belongs & I'll let you know the DPS. At ~32% crit * 1% miss = it might be as much as 0.32% more DPS!

But, I've never seen a single shred of evidence to suggest that the system you're talking about is used to determine hits/crits/resists for casters. Anyone have some?

 

Yes

And much better dpm (by 2+). And will partially resist less.

And will be less affected by latency!

Also, factoring hit is irrelevant: Every druid is going to be max hit (16) simply because of getting 4 from talents and 3 from totem -- the other 9 is REALLY easy to come by on spellfire. So multiplying crit chance by .99 or multiplying non crit chance by .99 does not make any real differance.

 

Efejel

I think you exaggerate the DPM delta. Also, if you're factoring JoW on the target I'm not sure there is a DPM delta, or that it doesn't favor Wrath.

With 1000 +dmg from gear/buffs, Wrath is 5.12 DPM and Starfire is 5.89 DPM--both before modifiers like CoS & Misery.
Will be with Veiled Noble Topaz. Actually, sadly, is not currently.

 

Kazanir

This is generally false. The majority of raid mobs have resist scores of 0 against various schools of magic. Partial resists seen as a caster are normally a level-based mechanic, not a resist score-based one. Curse of Shadows does nothing to influence the number of level-based partial resists you experience, and as such doesn't matter until you meet a mob that actually has some Arcane Resist (which is very rare at best.)

'War' is too small a word for what I'm fighting. Like a candle in front of the whole burning Sun. I told you. This is bigger than a war. Now, I am not going to die today. I have other projects, and other options.

You can come with me. I can protect you.

 

Boevis

Could you explain this part to me? I'm still working on statistics, but shouldn't it just be:
(d/3*q)+(2d/2.5*p)+(0*.01)
with q = (.99-p)
Since a miss doesn't reduce crit damage, it's easier to just reduce the chance to not crit by 1% and count the 1% miss as 0 DPS 1% of the time. And why not just reduce the cast time reduction and 2x damage together, since they both work off the same variable of crit%

 

Yes

Can you show me that math please?

 

Efejel

WRATH (10)
(422+475)/2 = 448.5 avg dmg before gear
1000 * .7386 = 738.6 contribution from 1000 +dmg
738.6 + 448.5 = avg hit of 1187.1
1187.1 / talented mana cost of 232 = 5.116 DPM


STARFIRE (8)
(605+711)/2 = 658 avg dmg before gear
1000 * 1.32 = 1320 gear +dmg contrib
1320 + 658 = avg hit of 1978
1978 / tal mana cost of 336 = 5.886

As for JoW--I'm afraid I don't have any information on the proc rate or if that rate is normalized by cast time. If we assume a return of 74 mana per proc:
WRATH (10) =
50% proc rate: 1187.1 / (talented mana cost of 232 - 37 mana avg JoW proc )= 6.088 DPM
75% proc rate: 1187.1 / ( 232 - (74 * .75) ) = 6.726 DPM
100% proc : 1187.1 / ( 232 - 74 ) = 7.513 DPM
STARFIRE (8) =
50%: 1978 / ( 336 - 37 ) = 6.615 DPM
75%: 1978 / ( 336 - (74 * .75) ) = 7.052
100% : 1978 / ( 336 - 74 ) = 7.545

So I was wrong about Wrath ever exceeding the DPM of Starfire at +1000 dmg, outside of Enhance Shammy debuffs.

 

Efejel

I think I switched the meanings of p & q in between defining them and using them, which probably wasn't helping matters! Don't worry, in the actual spreadsheet they're right. That I've checked 100 times.

Even without statistics, you can puzzle out the equation step by step. You have 4 possible scenarios (p = non-crit %, q = crit%)
1) The spell you are casting is not under effect of NG (the spell before didn't crit) and does not crit ((d/3)*pp)
2) The spell you are casting is not under effect of NG & does crit ((2d/3)*pq)
3) The spell you are casting IS under effect of NG & does NOT crit ((d/2.5)*qp)
4) The spell you are casting IS under effect of NG & DOES crit ((2d/2.5)*qq)

You have to model all 4 possibilities to get the accurate DPS make-up. If you apply the cast time reduction only when considering crits, you compound the value of crit. Say you're flipping a coin & every time you get 2 heads in a row, you get $5. You couldn't just take the chance of flipping heads & multiply it times $5. The same applies with crits--you can't just take the DPS done by a crit & assume NG is active. I made that same mistake the first time I tried to model NG procs and was horribly overvaluing crit.

(Also, my model never accounts for the "first cast" or any cast after a pause of more than 15 seconds. These should be insignificant variables over the course of 100+ casts though, and can be resolved by casting rank 1 moonfire on engage & after pauses of > 15 sec in real fights, if one is willing to assume that the GCD & ~25 mana used by rank 1 moonfire are taken care of by the regen of 15 sec of not casting and some sort of say, movement time to get back into position or to get into initial position).

 

Yes

Where does the constant of .7386 come from?

Also, your calculation method is really confusing me: I find it way easier to compute BASE spell damage modified by spellpowe and then added talents.

Ie: ( avg + spell x 1.2 ) *1.1 * misery x shadows

What I am getting for wrath constant is:

2/3.5 x 1.1 (35 pt talent ) * 1.1 (moonfury)

Also, multiply starfire dpm by 1.1 because cos is starfire exclusive.

Anyway, doing your wrath constant and multiplying by misery/critrate (1.27 )
I get
8.630316952 vs
6.832137873

Basically 2 dpm apart, as 1000 spelldamage is unreasonable and 1300 is more expected (bringing dpm to greater then 2 dpm apart)

Efejel, while YES, without all the modifiers the spell dpm is not 2 dpm off, it is 15% off, and adjusted for CoS and scaling it to critrate produces a 2dpm change!

 

Boevis

I believe he's using (2/3.5 + .1) * 1.1 as the wrath +dmg coef, and simply including that moonfury modifies the base damage in his 422-475 base damage, which I understand doing since he pointed out that moonfury modifies our base damage by something other than 10% (10.8% for wrath, 11.9% with strange rounding for starfire)

I would like some confirmation that the WoC modifier is added to the base spell coef and not multiplied, this is contrary to the observed behavior of our other Empowered talents (Rejuv and HT)

 

Efejel

I strapped on exactly 800 +dmg worth of gear on the char sheet (and no idol) with only my MotW (which I didn't let expire) and cast > 250 Wraths(10). The observed minimum was 1013, and the max was a 2130 crit.

Expected values using my equation: 1012.8 ~ 2131.7

I saved the SW stats segment, but I don't know if there's any way I can share it or whatever.

I'd challenge this assumption currently for non-top-end raiders, and post-2.1 with non-stacking consumables.

EDIT: Er, NM. See you throw in misery at the end.

 

Boevis

So the coef for wrath, and post moonfury base damages are different from what's expected. Before I keep going on my spreadsheet, could you confirm the following:

IS: .76 Coef
Moonfire: 369-432 (400 avg) initial, 726 DoT, 1126 total average. Initial .1524 coef. DoT .5158 coef

If the values for MF are different, I'll have to sneak in 2/2 Imp MF next time I respec for arena and see if it's Imp MF and/or Moonfury that are providing different results.

I suppose I'm going to have to separate talented from untalented (hey, some of us care what our DPS with a 27/0/34 spec is) And then there's the issue of Dreamstate HT specs only having 3 or 4 points in Moonfury.... /emo

 

Efejel

Spell pane for Moonfire(12): 366-429, 720 over 12 sec

Don't know coeff. for IS, but it's not modified by any talents. Believe it's (12/15) * 0.95 = 0.76... though this doesn't seem right. Time to go log in!

EDIT: 801 +dmg, 233/234 ticks, 6 ticks. 6 x 233 = 1398 - 792 (spell pane) = 606 / 0.76 =~ 797
EDIT 2: It's probably most appropriate to use the untalented values for determining the ratios of the coefficients?
( 305 + 357 ) / 2 = 331
331 / ( 331 + 600 ) = 0.3555 * ( 1.5 / 3.5 ) = .1524 (hey, that's where you got that from--the ratio determined with the talented number is diff)
600 / 931 = 0.6445 * ( 12 / 15 ) = .5158
As for testing Moonfire coefficients... :-/
I can give it a try. It's annoying as hell b/c the initial damage range is so large, and the mana consumption of spamming it.

 

Boevis

So IS is within rounding error of accurate, I'm happy with that.

I'm very much not happy with the WoW programming being seemingly incapable of rather simple math. That being said, I'll take a free 11 damage on starfire if they're going to give it to me.

Next time I respec (or whoever gets around to it), keep track of the spell pane point by point for Improved Moonfire, and Moonfury. I should probably also check Moonglow, I remember seeing discrepancies between base*.9*.91 and spell pane with a Moonglow/TS/GoN/Imp Rejuv.

 

Efejel

Went on test server.

702 +dmg from gear (according to the char sheet, but glove enchant of +20 dmg was only adding to +heal on char sheet--so maybe 722, but probably 702)

With 2/2 Imp MF and 0/5 Moonfury: 451 (902 crit) - 509 (1015 crit--sorry, it was boring) Tick: 265-266
[Spell pane: 335-393, 660]
With 2/2 Imp MF & 5/5 Moonfury: 492 (985 crit) - 555 (1110 crit) Tick: 289-290
[SP: 366-429, 720]

And for your various points in Moonfury problem:
From the spell pane, for 2/2 Imp Moonfire and Moonfury:
1/5 - 341-400, 672
2/5 - 347-407, 684
3/5 - 353-415, 696
4/5 - 359-422, 708

My apologies to the many, many, many, many Teromoths harmed during the collection of this data.

 

• Bluefish

No, it isn't. Some very fortunate Rogues/Warriors have enough good +hit gear to cap out, most do not, or aren't aware they should be trying to do so.

In the gear/buffs I had on our Gruul kill this week, iFF would have been 31 DPS for me. The other Rogue present also would have got 25+ DPS from it, the two tanks certainly weren't hit-capped, and we had a Fury Warrior chugging along too who certainly would have benefitted.

iFF also ups the aggro ceiling, noteworthy for a class with poor aggro reduction and no dumps. :P