Elemental Mechanics

Introduction

This document aims to highlight continuing mechanic issues for Elemental Shaman. It is a work in progress so it may change as new information is available. There may be some incorrect information as my Shaman is essentially “retired” which means I’m not storing the same amount of information in my memory as I used to.

Recent Blizzard Responses

On the WoW forums there was a response from Lore (from the CM team, and isn’t a developer) about why there hasn’t been much in the way of changes or discussion on Elemental recently.

So, part of the reason we haven’t had much to share about Elemental is that we don’t think there’s anything fundamentally wrong with the spec. Most of the changes you’ve seen to other specs in 7.2.5 thus far have been focused around addressing core mechanical issues. We’ve only recently begun the process of making changes intended to bring each spec’s performance more in line with the others.

The bulk of the feedback we’re seeing (in this thread and others) is that Elemental is fun to play, it just doesn’t deal enough damage in certain situations. That’s the sort of situation that we’ll solve through what we generally refer to as “tuning” – e.g, adjusting things like how much damage a spell does, the chance for an effect to activate, and so on. Based on the feedback, those sound like the sort of changes we’d consider for Elemental.“

The problem with this response is that there are some fairly glaring mechanical issues with Elemental, and in many cases these can’t be simply resolved by tuning hotfixes. With this post I’m going to run through as many different subjects as I can while explaining why each is an issue that needs a resolution.

Why Balance Tuning Isn’t The Answer

Part of the problem with Elemental is that there are many interactions between spells, stats & other effects that work in unusual ways. An example of this is the talent Ascendance, which removes the cooldown on Lava Burst for 15 seconds and increases its damage by your crit percentage. Many other classes get cooldowns that increase a stat like haste by a large percentage or an overall damage bonus to everything, either of which works with the “standard” rotation but just makes it do more. Using Arcane Mage as an example (which is my current “main” class/spec) Arcane Power increases damage by 30%, with a secondary effect of reducing mana costs by 30% too.

The value of Ascendance is derived from the difference between the standard rotation and pure Lava Burst spam, which is normally the damage difference between Lightning Bolt and the buffed Lava Burst. Flame Shock should be on the target already, maelstrom should be “dumped” before Ascendance to reduce the number of Earth Shock uses during Ascendance (only with the Swelling Maelstrom trait, otherwise it is not used), while Elemental Blast is the only other spell that should be used while Ascendance is up (if talented). The problem here is that if either Lightning Bolt or Lava Burst have their damage output tweaked the damage gain also changes, which will alter the value of Ascendance vs the other L100 talents (Icefury & Lightning Rod). In particular, if Lightning Bolt is buffed but Lava Burst remains as is then Lightning Rod gains in value while Ascendance loses value.

Overall, trying to balance Elemental via tuning tweaks to spells is like playing whack-a-gnoll: when you fix one problem another one occurs, which in turn causes other issues when you fix that one.

Rotation

Trade Offs Between Resource Spenders And Generators

For Elemental spells generally fall into three categories: maelstrom generators, maelstrom spenders, and a small group of non-maelstrom spells (eg: Elemental Blast). The key difference between Elemental and other specs is that some of the big damage abilities are also resource generators (eg: Lava Burst) while others are resource spenders (eg: Earth Shock). To further compound the problem Earth Shock is not a fixed resource cost but a “use all of your current resources” cost.

Due to the combination of both Lava Burst being a resource generator and Earth Shock being a variable cost resource spender the Elemental rotation runs into what I call a “decision conflict”.

Movement

The issue with Elemental movement is not so much the lack of damage but the large variation in damage, counter intuitive spell use, and requiring casting movement supporting abilities before moving.

Damage Variation

The two prime causes of this are Lava Surge procs and Earth Shock being a resource “dump” rather than a resource spended. For Surge, if you get one or more surge procs just as you start to move or while moving, it will contribute a massive amount to your damage in that short window. Similarly, if you’re not quite at your normal maelstrom spending threshold you can dump it all on Earth Shock, saving a stationary GCD for generator spell use. The difficulty is that Surge is a proc from each Flame Shock tick (and one of the few dot procs that aren’t tied to a single instance of said dot) so it’s not a reliable ability to use. Earth Shock is a one-and-you’re-done ability, with no charges or fixed cost to allow you to fill in a few GCDs of movement time.

Counter Intuitiveness

The “filler” ability for damage while moving is a somewhat surprising Flame Shock. This is because it deals more damage than Frost Shock, the only other instant ability without a minimum maelstrom cost or proc requirement. The problem is that when you use a 0ms Flame Shock it applies a 15 second dot which refreshes up to 19.5 seconds. If you have a 20ms Flame Shock up, lasting for 30 seconds and a 39 second maximum refresh, that longer dot is only refreshed up to 19.5 seconds. By casting a 0ms Flame Shock immediately after a 20ms one will result in losing nearly 20 seconds of dot time for zero DPS gain, and quite likely replace a dot buffed by Elemental Focus (10% damage bonus) too.

One of the core basics of caster DPS is “Always Be Casting”, covering such situations as “I’m casting Lightning Bolt and Lava Surge just proc’d. Should I cancel my current cast to use Surge?”, to which the answer is to follow the ABC’s. Side note: the reason why cast cancelling is a bad thing is because the DPET (damage per execute time) of your “big” spell is going to be a lot lower if you consider it against the time lost from cancelling the cast in addition to the actual cast time.

It’s counter intuitive because using Flame Shock to overwrite an existing dot is a net DPS loss if you end up wasting two or more dot ticks, which means it’s one of those rare times when doing nothing is a DPS increase.

Predicting Movement Requirements

One of the other interesting movement mechanics for Elemental is using a cast time ability to generate charges to use while moving, which requires using it before you expect to move but has a long enough cooldown that failing to use it at the right time can be costly. There are two of these abilities: the artifact ability Stormkeeper and the talent Icefury.

Stormkeeper is very much a case of getting the timing right as the charges it generates will be used on the next three Lightning Bolt casts, no matter whether you want to use them while moving or not, and will almost always be used as a damage cooldown rather than a movement ability. Icefury is a little more forgiving as it generates four charges for buffing Frost Shock damage, and thanks to not normally using Frost Shock for damage this means you can save the charges for a bit.

Not only do you need to predict movement so you can cast one or both of these but both abilities are fairly solid damage cooldowns as well, so if you save them for upcoming movement phases and then mis-time that pre-movement cast that’s more potential damage lost. This is especially problematic in PvP where things are more chaotic.

Itemization

Mastery & Stat Hardcaps

Caps on stats are when going over a quantity will make adding more of that stat worth less. Soft caps are when there’s a significant drop but it’s still a DPS gain, whereas hard caps are the point where that stat will produce zero additional gain. Critical Strike has an obvious 100% hard cap, with Haste having a 100% soft cap (where the GCD hits 0.75 seconds) and a 267% hard cap (where 2 second casts become 0.75 seconds). Versatility effectively has no cap of either sort.

The problem is with the hard cap for Mastery at 100% as Elemental Overload is a chance event similar to Critical Strikes. Thanks to the way that Mastery works the total amount of Mastery rating required to hit 100% is 14578, but when taking the Elemental Blast talent it causes a soft cap at 12178. The cap is so low vs the ~38k required for Crit & Haste because of how Mastery rating is converted into a percentage.

Back when Mastery was first introduced in Cataclysm it was converted from rating to a point system, and this was then multiplied by a spec specific value to return the percentage. It still converts this way but it has been streamlined to go directly from rating to percentage in the UI and you can work out the point to percentage conversion value from the base Mastery percentage divided by 8 (which is the default mastery quantity), so for Elemental it’s 2.25% per point (or 400 rating). Starting off at 18% is nearly one fifth of the hard cap, and the rating to percentage conversion is roughly twice as effective (400/2.25 = 178 per 1%) as for Crit (400) & Haste (375). Throw in a free 2400 rating from the Elemental Blast buff (you can get a second buff if it overloads, making this a more likely occurrence) and that’s 31.5% out of 100% without any gear at all.

If you look back at Warlords of Draenor you’ll remember that Overload was given to every class as a new stat called Multistrike which was removed for Legion. Elemental had a 20% base multistrike chance and a bonus to multistrike rating, so by the time Blackrock Foundry was over and Hellfire Citadel started players were hitting similar caps on Multistrike (usually with the Blackiron Micro Crucible which had a stacking multistrike buff). Even prior to that there were mastery cap issues related to that 100% upper limit, as I’m fairly sure the same issue existed in Mists of Pandaria too.

In Legion we’re rapidly approaching this point again. Using Notes as an example (a log of his is used later in this document) he has the following secondary stats:

Crit 10,352
Haste 3,882
Mastery 10,792
Versatility 2,780

This leaves just 1,386 rating until that soft EB cap with an average item level of 910, so with Tomb of Sargeras opening in June I expect we will start seeing players avoiding mastery heavy items to stay under the caps, which will be difficult as Mastery is one of the two best Elemental stats at the moment.

Further to this is the fact that players taking Totemic Mastery have a lower 76.5% cap for Lightning Bolt & Chain Lightning which translates to 10,400 rating, a value that Notes and many others are already over.

Stat Weights & Scaling

One comparison I’ve seen from players recently is comparing the stat weights of one class/spec against another, and then declaring that the spec with the lowest weights has “bad scaling”. There’s some truth to this but it’s also not a completely accurate comparison. Stat weights are a complicated feedback loop, with increasing the quantity of one stat having a positive effect on the value of other stats. You can see more information and some example graphs in my blog post on producing graphs showing the interactions.

For most specs every spell will have an interaction with Crit, Haste and Versatility, with Mastery either affecting all or some spells. Elemental is a unique case where Crit doesn’t interact with Lava Burst due to the auto-crit mechanic, but this is “countered” by having on-crit procs (Elemental Focus) and crit damage bonuses (Elemental Fury + some artifact traits).

Stat weight graphs for Elemental can be found here. The behaviour for Crit & Haste is as expected: Crit does not change in value until hit reaches around 83% where it starts diminishing, most likely due to refreshing Elemental Focus charges, and Haste has a slight negative feedback loop with itself. Note that the haste negative feedback loop is expected, as it happens with every spec I’ve analysed this way (Arcane/Frost Mage, Frost Death Knight).

Elemental - Mastery Weight Plot.PNG

Mastery, as shown in the above graph, is a little different. Normally for other specs mastery has no feedback loop, but we can see that around the 76% mark (ie: the EB + TM soft cap) there is a definite negative feedback loop happening. This feedback loop becomes greater the more mastery you have, as it increases the chance of getting two EB buffs (second comes from the overload) while also having Lightning Bolt going over a 100% chance (or rather, hitting that 100% hard cap because you a 110% chance of overloading doesn’t give anything extra).

One of the commonly referred to “proofs” as to why Elemental “scales badly” is that the stat weights for Elemental are usually lower than those of other DPS specs. The reason for this is that of all four secondary stats only Versatility affects all spells in an equal fashion.

  • Crit has almost no interaction with Lava Burst aside from slightly increasing the chance Focus will be consumed on a cast and granting a % damage bonus when using Ascendance.
  • Haste has a low interaction with Lava Burst as the cooldown is a fixed 8 seconds so the only increase in cast count is via Lava Surge.
  • Mastery may increase the cast frequency of Earth Shock through additional maelstrom generation but it has a limited effect. For every additional maelstrom generated per second the minimum count for Earth Shock increases by 0.6 casts per minute. This is an ideal situation and assumes no wasted resources, so as generation increases the chance of waste increases too. There’s also no interaction with Flame Shock damage or elemental pet damage either. Mastery also lacks any interaction with some artifact traits which have overload effects, namely Power of the Maelstrom (max 3 charges granting Lightning Bolt a 100% chance to trigger a second overload) and Static Overload (10% chance of triggering overload on all targets which effectively reduces Mastery chance by 10% for Chain Lightning)

There are intentional changes designed to counter some of this, including Elemental Fury which increases the crit damage bonus to 250% (without racials or further traits), although this is more of a chicken vs egg question as to which caused which. Making further tweaks to this may improve things for Elemental but ultimately the core problem for the spec is, for want of a better phrase, due to scaling issues which cannot be resolved without making mechanic changes.

If you look back to Warlords of Draenor you’ll see that there was both a multistrike “attunement” (granting 5% more rating) which also granted a passive base chance of 20%. Multistrike was the best Elemental stat at the time, and that extra 20% was because it was necessary to keep that overload feeling from the previous two expansions. The problem here was that capping out multistrike was possible, and Elementals power relative to gear was higher at low gear & lower at higher gear. I can’t remember where my source data for this is, but it was something I identified at the time.

In Legion & Warlords the best stat for Elemental was chance based while also having attainable hard caps, and in both cases this would have a significant impact on DPS output.

Proc Effects

Ignored

There are a number of charge based proc effects that are effectively ignored, playing no part in altering the rotation to any extent. These include Power of the Maelstrom, Stormkeeper charges and Elemental Focus, the latter of which will occasionally be used to specifically buff Flame Shock. The primary issue is that the charge based nature essentially just requires all charges to be used before expiring rather than altering the rotation priority which is what a timed buff could do.

One common question from players during the Legion beta was around whether these charge procs would alter the rotation, as getting extra power for an ability seems like a good reason to use that first over something else. For a while we considered a “Focus Weaving” approach when Elemental Focus was a 20% (or 40% with T19) damage bonus and Echo of the Elements had 3 charges for Lava Burst (which allowed players to delay spending a charge without DPS loss) but changes to all of these were made, which I partially suspect were intended to stop this potential behavior.

It’s also possible to overwrite existing charges with a new set for those that proc off regular events (Power of the Maelstrom from Lava Burst casts, Elemental Focus from direct spell crits) but even then losing charges is “better” DPS than trying to take advantage of them. Stormkeeper is slightly different in that the player triggers the buff generation, but because the buffs still don’t alter the rotation this ability is pushed down to the end of the rotation priority to decrease the chance that less than three Lightning Bolt casts will occur before it expires.

Lava Surge is an example of a “good” proc, where you want to know it happens so you can react accordingly. The “bad” ones are those which could be removed entirely from the game without making an observable difference to the player aside from less damage.

Underwhelming

Similarly to the above, procs like Volcanic, Fury of the Storms and Seismic Storm are at least visible some of the time, but still fail to contribute much in the way of reactions from the player. For Volcanic it can be slightly frustrating to have your target move and see that a little volcano effect is sitting where they once were. Seismic Storm has a low proc rate with high damage.

The summoned Lightning Elemental for Fury of the Storms looks cool, but it’s triggered off using Stormkeeper and has low damage/duration so it’s output is fairly small.

Player Control vs Randomness

Writing about randomness would essentially require re-writing an article by Bloodmallet, so instead I will just link to the post. It is worth reiterating the point that low chance/high impact procs like Static Overload or Seismic Storm are the cause of fairly large swings in DPS or resource generation. Ending up on the low end of these can feel bad because you’re hoping something will proc and it doesn’t, but on the high side it can be equally as bad when you cap out maelstrom.

Fire Elemental as a Cooldown

Cooldown Power

The closest thing Elemental has to a cooldown ability that isn’t either tied to the artifact weapon or a talent is Fire Elemental, which summons a pet for 1 minute with a 5 minute cooldown. As a cooldown its burst potential is relatively weak, only gaining power from lasting a longer time than other cooldowns.

Looking at a Mythic Krosus log from Notes (https://www.warcraftlogs.com/reports/JRfB2h9NZcjYT4z7#fight=19&type=damage-done&source=21) we can see he gets two uses of Fire Elemental for a ~70 second uptime and roughly 81.8k DPS while active or 25.9k dps over the course of the whole fight. By comparison Notes total damage to Krosus (excluding adds) is 741.6k DPS, so as a cooldown Fire Elemental feels fairly weak at just 3.5% of the total damage, even if it is 5.2 million damage in a single keypress. This is a similar problem that occurred with the now removed Searing Totem, which was a low DPS totem but with a 60 second duration so the DPET of the spell was reasonably high but the actual DPS output was low.

Cooldown Reduction Problems

This feeling of “lack of power” compounds with the Elementalist trait (reduces Fire Elemental CD by 2 second for every Lava Burst cast) and the upcoming T20 2pc (Fire Elemental cooldown is reduced by 3 seconds every time Flame Shock crits) and 4pc (Flame Shock always crits while Fire Elemental is active) bonus. The T20 bonuses are limited to triggering one reduction every second, so there’s a maximum of 180 seconds off the 300 second cooldown while Fire Elemental is up. Elementalist itself can shave ~90 seconds off by itself, but these effects don’t stack additively together so you have to look at it on a CD reduction per real time second.

Sticking with Notes log we can estimate Lava Burst is cast once every 5.8 seconds, or every 1.5 seconds during Ascendance. This means Elementalist reduces the cooldown by roughly 1.34 seconds per real time second or 2.33 seconds with Ascendance, while the tier bonus has 40 FS ticks to give ~3 seconds reduction per second. This combines to reduce the cooldown by 215 seconds in the first minute (Asc: 1.33 * 15s, T20 FE: 2 * 60s, Elementalist: 0.34 * 45s + 60). After this the reduction is 2.54 seconds (0.34 + 1.2) per real second suggests the cooldown will end roughly 52 seconds after it expires for a total cooldown of 112 seconds. That’s a 63% uptime on Fire Elemental but we can see from Notes log that he had a 28.7% uptime with just Elementalist on its own. With 5.2m damage per cast this gives a current average DPS of 24.9k or 46.6k with the set bonus, which is a ~6% damage increase when compared to the 741.5k damage Notes dealt to Krosus.

In Summary

Elemental does indeed have core mechanical issues. These contribute to having a higher skill requirement for low level players, while at the same time having fewer opportunities for high level players to shine. Rather than the “easy to learn, hard to master” ideal I’d say it’s “slightly difficult to learn, nothing much to master”. Simply increasing damage numbers will plaster over these core issues, rather than making any meaningful change.

Advertisements

About binkenstein
I'm a geek, living in Christchurch, New Zealand.

2 Responses to Elemental Mechanics

  1. Mattrik says:

    Thanks Binkenstein, so glad to have you still assessing these things.

  2. Noshock says:

    Holy Jezuz Bink if Blizard would finally hire you, shaman would be a much better spec! Don’t lose hope! Keep on spreading out the message for the sake of the elemental shamans! Thank you so much!

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s

%d bloggers like this: