Moonkin Raiding - Page 14

dukes · 12/25/08, 3:06 PM

My cast time (according to Quartz) is ~0.9 seconds base, ~0.6 seconds with NG while under bloodlust. This is with 470 haste, 3/3 CF and normal raid buffs.

Starfox · 12/26/08, 1:11 AM

WoA Totem, CF, ImpMK,
1.5/(1.05*1.03*1.03) = 1.346566
+BL
1.5/(1.05*1.03*1.03*1.3) = 1.03582
3.58% haste (118 haste rating) from gear and your wrath starts to bang against the gcd.
It will be hard to stay under 118 haste rating with rading gear

mokg · 12/26/08, 3:46 AM

Sorry I should have said a significant amount. Fact is you should have ~50% chance to crit with wrath, that means a few seconds to proc 30% crit for 15. If you don't want to do it you don't have to, I'll enjoy doing ~5700 on PW because I choose reality over inaccurate spreadsheets/math.

I could go into a debate on how crit is better then haste(patchwerk may be one of the few exceptions to this, but patchwerk isn't everything to most people), but I'll spare you math nerds who will stand by unrealistic averages over uncertain lengths of time fighting to the death over your calculations.

dukes · 12/26/08, 6:59 AM

Originally Posted by mokg

I could go into a debate on how crit is better then haste(patchwerk may be one of the few exceptions to this, but patchwerk isn't everything to most people), but I'll spare you math nerds who will stand by unrealistic averages over uncertain lengths of time fighting to the death over your calculations.

As much as I prefer not feeding trolls, I would actually quite like to get into a debate on this. The value of haste has been bugging me - I don't really go all out to stack it (I prefer split crit/haste items), but I'd like to put a value on each in real situations. I've always preferred the "real" approach over the theorycrafting one, as sometimes theorycraft just doesn't add up due to having to simplify particular mechanics.

Crit provides:
- Double Damage on Starfire/Wrath crits (and starfall/hurricane/typhoon in AoE situations) on a RNG powered system
- Mana return
- 0.5 seconds reduction of cast time
- Better/faster eclipse proc rate

Haste provides:
- A guarenteed faster casting time on all spells until the GCD cap is reached (effectively unattainable for all spells other than Wrath with NG, unless under bloodlust).

You need ~46 crit rating per 1% crit, while you need ~33 haste rating per 1% crit, which makes haste 1.4 times more valuable in pure percentage terms than crit. The question is, do the actual in game benefits make crit 1.4 times more valuable than haste?

One of the issues with haste is that there is a break-point for each fight and eclipse rotation as to where a specific amount becomes more valuable. For example with eclipse there is a point where x amount of haste will allow you to cast one more spell within the eclipse. Obviously this is hard to model due to the expected number of crits providing NG, which means there's actually a lot of break points dependant on how many crits you get. One of the fights where I find this has most of an effect is Heigan, as once you get to enough haste you can easily weave wrath casts in between movement.

It is also true that most fights are not long enough for the randomness of crits to balance out. Haste meanwhile provides a very real exact benefit to all spells all the time.

Unfortunately I don't think I can actually put a value on which is better, but from experience I prefer haste over crit in those instances where I have to choose. I still put spellpower as my highest priority though.

Trouble · 12/26/08, 10:24 AM

Originally Posted by dukes

One of the fights where I find this has most of an effect is Heigan, as once you get to enough haste you can easily weave wrath casts in between movement.

I've focused mainly on crit itemization and even with my low amount of haste I can get off one wrath between every movement during the dance.

But yeah your question is a good one and has been my main one with regards to itemization. We generally have a choice between haste and crit on items at the current tier of itemization so pinning down what we should be going after is key to maximizing our DPS. I assume as we increase in tiers the gear will become better and begin feature both crit and haste, but until then we have choices to make.

• Hamlet · 12/26/08, 5:43 PM

Originally Posted by dukes

As much as I prefer not feeding trolls, I would actually quite like to get into a debate on this. The value of haste has been bugging me - I don't really go all out to stack it (I prefer split crit/haste items), but I'd like to put a value on each in real situations. I've always preferred the "real" approach over the theorycrafting one, as sometimes theorycraft just doesn't add up due to having to simplify particular mechanics.

Crit provides:
- Double Damage on Starfire/Wrath crits (and starfall/hurricane/typhoon in AoE situations) on a RNG powered system
- Mana return
- 0.5 seconds reduction of cast time
- Better/faster eclipse proc rate

Haste provides:
- A guarenteed faster casting time on all spells until the GCD cap is reached (effectively unattainable for all spells other than Wrath with NG, unless under bloodlust).

You need ~46 crit rating per 1% crit, while you need ~33 haste rating per 1% crit, which makes haste 1.4 times more valuable in pure percentage terms than crit. The question is, do the actual in game benefits make crit 1.4 times more valuable than haste?

One of the issues with haste is that there is a break-point for each fight and eclipse rotation as to where a specific amount becomes more valuable. For example with eclipse there is a point where x amount of haste will allow you to cast one more spell within the eclipse. Obviously this is hard to model due to the expected number of crits providing NG, which means there's actually a lot of break points dependant on how many crits you get. One of the fights where I find this has most of an effect is Heigan, as once you get to enough haste you can easily weave wrath casts in between movement.

It is also true that most fights are not long enough for the randomness of crits to balance out. Haste meanwhile provides a very real exact benefit to all spells all the time.

Unfortunately I don't think I can actually put a value on which is better, but from experience I prefer haste over crit in those instances where I have to choose. I still put spellpower as my highest priority though.

The model in Adoriele's spreadsheet favors haste by a good bit, which I think jives with most people's experience.

If we just want to try some envelope math, though:
Note that a crit does 2.09x damage and (effectively) takes 5/6 as long as a hit (for Starfire). So if you're critting at 50% (seems typical for a raid), 100 casts gives 50 crits and 50 hits, doing 154.5 damage in 91.7 time. Going up to 51% crit makes you deal 155.59 damage in 91.5 time. A DPS increase of about 0.89%.
1% haste increases your DPS by 1/(1+H), where H is your total haste% from rating. So 0.87 to 0.9% for typical gear ranges.

The large difference in the value to your Starfire spam should be enough to decide the question--haste is going to be about 1.4 times as effective. The numbers will be slightly different for Wrath spam, due to NG clipping, but not nearly enough to tilt the scale when you're casting Wrath well under 1/3 of the time.

----------

The effect from shortened Eclipse cycles will be of yet a smaller order of magnitude. Again assuming 50% crit, 1% more increases your chance to Eclipse with Wrath from 30% to 30.6%.

Assuming chance E to proc Eclipse, the chance of a proc on precisely the nth cast is:
$E(1-E)^{n-1}$ .
The expected number of casts to proc is then:
$\frac{E}{1-E}\sum_{n=1}^{\inf} {n(1-E)^n}$ ,
Which converges to 1/E.*

So the 1% crit gives 1/.306 - 1/.300 = 0.06 fewer Wrath casts per cycle. On the order of a tenth of a second in a 30+ second cycle, which will be dwarfed by the gains in Starfire spam DPS discussed above.

Also, 1% haste would save 1/.3*0.01= 0.03 effective Wrath casts anyway.

*I'm handwaving this for now. It would be cool if someone could remind me how to prove it.

Starfox · 12/27/08, 12:05 AM

Originally Posted by Arawethion

Assuming chance E to proc Eclipse, the chance of a proc on precisely the nth cast is:
$E(1-E)^{n-1}$ .
The expected number of casts to proc is then:
$\frac{E}{1-E}\sum_{n=1}^{\inf} {n(1-E)^n}$ ,
Which converges to 1/E.*
*I'm handwaving this for now. It would be cool if someone could remind me how to prove it.

the math

E .. Chance to procc eclipse = Crit% * 0.6
Chance to procc on cast
$1: E$
$2: E(1-E)$
$3: E(1-E)^2$
$n: E(1-E)^{n-1}$
expectation value: $Wrathcasts = \sum\limits_{i=1}^{\infty}Ei(1-E)^{i-1}$

Multiplying every summand with $1-E$ gives an "easy" to solve geometric series, and dividing the that by $1-E$ gives us the EV from our series
$Wrathcasts =\frac{1}{1-E} \sum\limits_{i=1}^{\infty}Ei(1-E)^{i-1}(1-E)$

$Wrathcasts =\frac{1}{1-E} \sum\limits_{i=1}^{\infty}Ei(1-E)^{i}$

$i=0 \rightarrow E\cdot0\cdot(1-E)^0 = 0$

$Wrathcasts =\frac{E}{1-E} \sum\limits_{i=0}^{\infty}i(1-E)^{i}$

With our everyday mathskill we solve that with the widely known $\sum\limits_{i=0}^{\infty}i(q)^{i} = \frac{q}{(1-q)^2}$ :P
$Wrathcasts =\frac{E}{1-E} \frac{1-E}{(1-(1-E))^2}$

$Wrathcasts =\frac{E}{1-E} \frac{1-E}{E^2}$

$Wrathcasts = \frac{1}{E}} = \frac{1}{Crit\%\cdot0.6}$

Woodlum · 12/29/08, 5:55 PM

Is there a macro or some way to place targeting circles and cast the associated spell with a single button push; i.e. hurricane or treants, on top of yourself or in any given static position... say like 30 yards to the front? I really hate having to bother with using the mouse for that and I feel like it is costing me cast time and it also feels like a hinderance to my movement.

dukes · 12/29/08, 6:29 PM

You can pre-cast the spells e.g. start a starfire cast, hit the button for hurricane/treants and position it, then actually cast it ~0.2 seconds before starfire ends. You then have a continuous stream of casting, instead of having to wait for the previous cast to complete.

Tarranus · 12/30/08, 4:32 AM

All the balance threads are starting to meld into one another, but out of curiosity. Assuming you are now an eclipse moonkin, and your SF-eclipse has just faded. What do you cast in the intertemporal period where u cannot proc another eclipse, SF correct?

spartakos · 12/30/08, 5:07 AM

It depends on your mana and the duration of your moonfire dot. If both are covered go for wrath.

Trouble · 12/30/08, 11:21 AM

Originally Posted by spartakos

It depends on your mana and the duration of your moonfire dot. If both are covered go for wrath.

Are you sure about this? I would think Starfall is what you want to be casting as often as possible due to not clipping the GCD. I would also think your Idol choice factors into this. If you have been able to get the Starfall idol you would have even more incentive to cast Starfall.

Humbaba · 12/30/08, 12:09 PM

Originally Posted by Trouble

Are you sure about this? I would think Starfall is what you want to be casting as often as possible due to not clipping the GCD. I would also think your Idol choice factors into this. If you have been able to get the Starfall idol you would have even more incentive to cast Starfall.

You mean Starfire, right?

• Adoriele · 12/30/08, 12:21 PM

Originally Posted by Trouble

Are you sure about this? I would think Starfall is what you want to be casting as often as possible due to not clipping the GCD. I would also think your Idol choice factors into this. If you have been able to get the Starfall idol you would have even more incentive to cast Starfall.

s/fall/fire/g, but yes. Although generally you're going to want to use the idol the matches your main nuke (unless you happen to have Raven Goddess, which is still better in Naxx for raid DPS contribution). At end Naxx, even assuming no penalty on Wrath for clipping the GCD (as relates to spell queueing, you're still capped at 1s cast anyway) your Starfire spell will have a higher DPET than Wrath in all cases (though it's close, with about 30 DPET difference in my default end-Naxx set). Starfire does scale better, though, so your mileage may vary with lower levels of gear.

If you'd like to double-check which is the better nuke for you, the DPET calculations in WC should give you a good idea. If you'd like to check scaling numbers, set your cast cycle to SF Spam and W Spam and check the scaling calcs on the first page (you should probably remove Eclipse from your talent spec temporarily, as that will give you merged results). Interestingly, it looks like they scale almost exactly the same with Spellpower, though of course Wrath scales much less with other stats, especially with Haste where it gets less than half the benefit that Starfire does. Also obviously it scales much more poorly with Latency.

dukes · 12/30/08, 12:38 PM

Originally Posted by Adoriele

(unless you happen to have Raven Goddess, which is still better in Naxx for raid DPS contribution).

Correction, better in 25 mans. In 10 mans, assuming a non-stacked group, then the Shooting Star idol will still be better, although any mage will probably love you for using Raven Goddess (crit scales insanely well for a deep fire FFB mage).