Elemental, Opinion, Shaman, Theorycraft, WoW

Elemental: Waiting for Lava Burst

I remember back in the WotLK beta when the shaman community theorycrafters were trying to make sense of this new Lava Burst ability.  At first, it seemed like the spell not only vastly reduced the value of our beloved crit rating, but the amount of damage it did combined with its cooldown and fixed duration of Flame Shock meant it interacted in a really funky way with haste.  And not the good kind of funky with disco balls and flashing lights and guys in white suits contesting dance offs.  I’m talking about the bad kind of funky, with the several-weeks-old raw fish and students’ sock drawers.

Haste, it was thought at first, would only be valuable if it allowed us to perfectly fit a certain number of Lightning Bolts into 8 seconds (the cooldown of Lava Burst).  Charts were constructed showing the value of haste as essentially nil until that magic point, upon which damage shot up because of the extra spell fitting in.  Of course, being pretty smart folks (smarter than me), the shaman community figured out that this was Wrong™ – not least because of forgetting about the second part of the cycle where you had to recast Flame Shock.  The assumption and error, however, have continued among the less well informed right up to the present.

The question is: Is there a point at which it’s worth not casting an extra Lightning Bolt and instead waiting for Lava Burst to come off cooldown?

Disclaimer: this post is tagged “opinion” for a reason.  While I believe my own findings are consistant with that of the broader shaman community, I didn’t put in enough time or effort to really definitively and entirely address the full range of possibilities.

What I did do was adjust Bink’s SEIC spreadsheet to compare a rotation that always cast Lightning Bolt until Lava Burst had actually cooled down to a rotation that always filled cooldown gaps of less than LB’s cast time with nothing, that is, with waiting for Lava Burst to cooldown.  It’s something I’ve played with for several days using different figures, but I finally used a specific set of variables and plugged them into a graph.

How I tested

Both tests used figures of zero latency, 3500 spellpower and 45% crit chance, with glyphs of Lava, Lightning Bolt and Flame Shock and no Totem equipped.  The focus was on the first part of the rotation, that is, the part that doesn’t include recasting Flame Shock (Note: when I did test earlier in the week using both parts of the rotation, the differences were more noticable; however the method I used was even less reliable and I didn’t save the results).

First Test

The first test used no tier bonuses and haste increments of 1%.  As you can see, the values start off pretty close and then diverge dramatically for a while.  As 27% haste is approached, they converge again until the Wait DPS actually exceeds the No wait DPS by 11 damage per second.  They then rapidly diverge again.  Click the picture for the full graph.

Click for full version.

Second Test

The second test used the same variables but four pieces of tier 9 for the extra Lava Burst damage.  The results are almost the same except when Wait passes No wait, it’s by a massive 14 DPS at peak and hangs around the same sort of value for a couple of extra haste percentiles.  The gain is less than 0.2%.  By contrast, the biggest difference between wait/no wait is nearly 6.5% in favour of no-wait.

Test conclusions

First, let me again stress that this is not definitive – it is merely suggestive and may or may not be indicative of reality.

The basic conclusion is that there are potentially situations where it is theoretically slightly higher DPS to wait for Lava Burst to cool down.  However, these situations occur at very specific haste values (and these much lower than the amount of haste most raiding elementals will have at present).  And the DPS increase resulting from waiting in that window is tiny, whereas the general DPS increase from not waiting tends to be huge.

I just thought this picture was amusing.

In reality

Now factor in latency, encounter dynamics (e.g. movement), haste variables (e.g. trinket procs, having/not having a moonkin, etc.) and player error.  Then after that, factor in all the usual random noise associated with DPS (variations on how often you critted, procs, circumstances alternately favouring you or hindering you).

The result is that there’s absolutely no freakin’ way you can know if it’s really better to wait for Lava Burst to cool down or not even if with your particular gear setup a spreadsheet or simulation says it is.  And if you do wait and get it wrong, then you’re losing a lot more DPS than you would have gained by waiting and getting it right.


My opinion, which seems to be borne out by the data, is that it’s simply not worth waiting for Lava Burst to cooldown – it’s always better to chuck in another spell, either a Lightning Bolt or a Chain Lightning.  If you end up with a really really small gap, like 0.1 seconds or something, it might well be better just to wait, mash your Lava Burst key and hope for the best!

More information

Both Rawr and Bink’s spreadsheet attempt to work out whether it’s better to wait or not, and then return the appropriate rotation.  You can play around with these tools yourself if you want to investigate further.

Tomorrow I’ll take a closer look at the use of Chain Lightning in a single-target situation.



3 thoughts on “Elemental: Waiting for Lava Burst

  1. Nice post. I can say personally when I’m raiding I will try to fit in an extra CL or LB depending on movement required in the fight. I just found your blog and from the first post I will definitely keep reading.



    Posted by Hakawa | October 26, 2009, 10:33 pm
  2. Yes, the post is year-old but I’m still quite sure that “gearing for X LB’s between LvB is a way to go”, and for me it’s really clear that the chart is quite set-up to prove you point while ignoring the reality, because there’s a big error in it.

    The thing is, if you are gearing for specific haste amount, then after reaching it you don’t stop getting ANYTHING AT ALL, you just start getting OTHER STATS instead of haste – and your chart basically ignores that. I’ll explain.

    Let’s look at 25% haste. When we go right, we see one guy (“no wait”) going above. Why? Because “wait”-guy doesn’t get benefit from the extra haste (he waits), but in reality he’s not stupid and won’t stack up extra haste, since he knows it will go waste. He would be getting the haste of “no-wait”-guy in crit or spell power. So when “no-wait”-guy had 35% haste, “wait”-guy would have same 25% haste, with a lot more spell power and you should actually compare the “no-wait”-guy with 35% haste and X sp to “wait”-guy with 25% haste and Y sp.

    That’s the idea of “waiting”, more making sense as a “gearing for specific amount of haste”.

    Posted by someguy | October 3, 2010, 4:21 pm
    • This post is not about “gearing for X LBs” or even haste in general. It’s addressing whether it’s ever better in real-terms to cast nothing for a short period in order to cast Lava Burst sooner, and the answer is “no”. If you had geared for a specific amount of haste then presumably you would not have any such delay and this would be a non-issue.

      Gearing for specific amounts of haste is covered very extensively in other articles, and the line of thinking you pursued in your comment is indeed very solid – except when you plug actual numbers into the argument, it leads to conclusions contrary to the one you reach. But as I said, that is not what this article is about. And it’s all changing in a week or two anyway 🙂

      Posted by Charles | October 3, 2010, 5:40 pm

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Get emailed about new posts!

Join 27 other followers

We write about

%d bloggers like this: