Thursday, October 9, 2008

Why Eclipse is Still Bad (for PvE)

For those of you that missed it Ghostcrawler graced us with her presence on the WoW Beta forums to update us on how they planned to buffing Eclipse to make it more attractive. Here is what she had to say.

It now has a longer cooldown (40 sec) and a shorter duration (10 sec) but the effects on Wrath and Starfire are doubled. It is overall a buff to the damage to justify the proc-watching aspect. We also hope the longer cooldown and shorter duration will mean it takes less of your bandwidth to do so.
This comment is a little confusing. Several people have asked if this is actually a buff or really a nerf. The math gets a little complicated, but it is indeed a buff over the old version. However, that does not mean it is a good talent or that PvE moonkin should pick it up.

Please let me show you why Eclipse is still a bad talent.

Possible uses for Eclipse:
There are two possible ways to use Eclipse. I call them the Opportunistic method and the Precast method.

With the Opportunistic method you use your normal casting rotation until you get an Eclipse buff, and then you switch to the buffed spell while you still have the buff.

With the Precast method you start casting your secondary spell with the intent to buff your primary spell.

Stat Assumptions:
Spell Power: 1800
Haste: 9.17% including Celestial focus and Imp MA
Hit: Capped
Crit SF: 37% - Includes Imp Scorch, MA, Nature's Majesty, and Imp IS (not unreasonable assuming a 15% base from stats)
Crit Wrath: 34% - Includes Imp Scorch, MA, Nature's Majesty
Talent build: Link

Standard Starfire DPS:
First off lets look to see what the DPS is for the standard IS, MF, SF*x rotation using these stats and assuming all three spells are glyphed.

IS Damage = (((1290 + (1800 * 0.76)) * 1.09) * 1.3) * (14/12) = 4,394
MF Damage Tick = ((800 + (1800 * 0.52)) * 1.29) / 4 = 982
SF Damage = (((1120 + (1800 * 1.2)) * 1.19) * 1.37) + 982 = 6,329

Starfire has an average cast time of 2.579 seconds, therefore you will cast Starfire 5 times before you refresh Insect Swarm. Therefore, Starfire and the IS, MF, SF*x rotation will have the following DPS.
Avg Starfire DPS = 6,329 / 2.579 = 2,454
Avg Rotation DPS = (4,394 + (5 * 6,329)) / (1.374 + (5 * 2.579)) = 2,526
The Opportunistic Method:
Update: A new possibility has been brought to my attention on how to make this method work, and it looks pretty good. I have posted a new analysis here.

In previous posts I've established that Starfire is the primary Raid DPS spell in WotLK. So, to use the Opportunistic Method we would start casting our normal Starfire rotation, then switch to wrath if and when we got an Eclipse buff.

It is fairly simple to understand, but lets see why it doesn't make sense mathematically.

For the Opportunistic Method to work, an Eclipse Buffed Wrath will have to have a rotation DPS greater then 2,454. We can calculate Wrath's DPS with eclipse as follows.
Avg Wrath Damage = (((588 + (1800 * 0.67))*1.39) * 1.34) = 3,342
Avg Wrath DPS = 3,342/ (1.5 / 1.0917) = 2,432
As you can see an Eclipse buffed Wrath still does less damage then the standard Starfire. Therefore, there is no reason to switch from the standard Starfire rotation even if Wrath has a 20% buff.

The Precast Method:
With the precast method we would start casting Wrath to receive the Starfire buff. Switch to Starfire as soon as you get the Eclipse buff and stick with Starfire until the cooldown is up.

First, we need to determine how much precasting it will take on average to get the Starfire buff. As stated in the assumptions Wrath has a crit rate of 34% in this example. On top of that each Crit has only a 60% chance to proc Eclipse. So each Wrath cast has 20.4% chance to proc Eclipse. This means that on average you will have to cast Wrath 4.90 times before you get the Wrath buff. On top of that it is reasonable to assume that we will not recognize that we have the buff right away. Therefore, lets assume that one Wrath cast occurs during the buffed period. This brings the total number of Wrath Casts up to 5.90.
Wrath's average cast time and damage output are:
Avg Wrath Damage = (((588 + (1800 * 0.67)) * 1.19) * 1.34) = 2,861
Wrath Cast time = 1.5 / 1.0917 = 1.374
Therefore, during this precast period you will on average do 17,334 Damage over 8.11 seconds, and it will cost 1,420 mana.

With the Eclipse buff Starfire's average cast time and damage output are:
Avg Starfire Damage = (((1120 + (1800 * 1.20)) * 1.19) * 1.67) = 6,518
Starfire Cast time = (3 - (0.5 * 0.67)) / 1.0917 = 2.441
The next question is how many buffed Starfire's can we get off during the buffed period. I have already indicated that we will lose 1.374 seconds of the buff due to slow reaction times and casting a Wrath during the buffed period.

Theoretically you can get 3.53 Starfire casts off during the buff period but parcial casts will not receive the buff. Therefore we should round down to 3 Starfire Casts.

So, during the buffed period with Eclipse you will on average do 22,502 damage over 7.32 seconds, and it will cost 794 mana.

If you combine this with the pre casting period you do a total of 39,836 damage over 15.43 seconds at the cost of 2,214 mana. This means that if you precast you will have an average DPS of 2,582.

If you compare this to the average DPS of the standard IS, MF, SF*x rotation of 2,526 you can see that the Eclipse can result in a 2.22% increase in DPS while Precasting. However, this does not consider the cooldown.

With the cooldown you shouldn't precast until 40 seconds after you receive the buff. Then on top of that it will take an average of 6.74 seconds to renew the buff. Meanwhile, we can only get 3 buffed starfire's off meaning that the buffed period only lasts 7.323 seconds.

So, the buff will be applied once ever 46.74 seconds and it will be useful for 7.323 seconds. This means that the Eclipse buff will only be present 15.67% of the time.

Therefore investing 3 points in Eclipse will only buff a moonkin's DPS by 0.349%.

1. Opportunistic casting Wrath casting will actually reduce your DPS on average. Therefore, there is no reason to switch Wrath with the Eclipse buff, unless there is some fight mechanic that makes Starfire less desirable.

2. Using the precasting method can result in a small increase in DPS. However, it requires quite a bit of micromanagement. It would be quite easy to mess up the rotation and lose some if not all of your gains.

3. Ultimately, Eclipse just isn't worth the points for a PvE Moonkin. It is so random that our performance could see significant changes from fight to fight even if played perfectly. On top of that there are several talents in the Balance tree that are regularly passed up but provide a similar increase to DPS with much less hassle.

Note: I don't know much about it, but apparently PvP moonkin are quite happy with this change. The talent is much more attractive with Wrath as your primary spell, and apparently using the buff in conjunction with the PvP set bonus can result in significant burst damage.


Eric said...

Some great math you've done here. It's nice to have a another Moonkin by my side that cares =) Keep it up!

Anonymous said...

It amazes me that they are still trying to cram this POS talent down our throats. About the only person I have ever seen consistently saying that its worth taking is GC. I would personally love to see how she plays if she thinks it is...

Thorn said...

Thanks for the maths work. It confirms what little testing I was able to do on test dummies. Have you looked at Owlkin frenzy? I like it for leveling but suspect in most raid situations it won't proc enough to be worth justifying.

Anonymous said...

Wow, thanks for doing all of the math to prove your point. I've been questioning this talent since it was mentioned but I can see why it would be nice in PVP with the right set bonus. It's still a POS for PVE in my opinion.

Cdin said...

Thanks everyone for reading and making comments.

This is going to sound strange but I'm about to write new post that is actually in favor of putting a point into Eclipse. Erdluf, over on the Moonkin Repository brought up a new way to use Eclipse that actually proveds a nice DPS increase. Look for it soon.


Owlkin Frenzy is really hard to quantify so I haven't really tried.

My sense is that it isn't a horrible place to spend a few points. In fact, I plan on putting a full 3 points in it on Tuesday when the patch drops because BT and SWP have lots of raid wide damage that will make it worth while.

The reports I'm hearing out of the beta indicate that quite a few of the fights in Naxx also have a lot of raid wide damage so it may be worth while there also.

Siobhann said...

Thanks for doing all that analysis. I'm going to be leveling moonkin and it is great to know which talents are good.