Theorycraft 101: Spellpower, part 2

Spellpower's effect on DPS is very simplistic in nature. All it does is increase the average damage of a spell by a proportion, which is nothing more than the coefficient. Say you are a mage and your Fireball does 100 damage. Since Fireball has a base cast time of 3.5 seconds, it has a 100% spellpower coefficient. Therefore if you have 100 spellpower, your Fireball will now do 200 damage instead of 100. In this respects, spellpower is very binary.
The mistake that many players make is assuming that higher coefficients equate to higher spellpower scaling. This is not true. All spell coefficients are completely normalized so that each and every spell in the game scales from spellpower at exactly the same ratio. To view a spell's actual spellpower scaling, however, you do not look at the coefficient but rather at spellpower per second (or SPS). Determining a spell's SPS consists of a very simple function: divide the spell's coefficient by its cast time. For example, Wrath has a base cast time of 2 seconds, giving it a coefficient of 57.14%. Lightning Bolt has a base cast time of 2.5 seconds, giving it a coefficient of 71.43%. Fireball has a base cast time of 3.5 seconds, giving it a coefficient of 100%. Although their coefficients are different, all three spells scale exactly the same based from spellpower. For all of these spells, their SPS is 28.57%. (Note that if you divide each of the numbers provided, you may notice some slight inconsistencies after the fourth decimal point. This is merely an account of rounding. When using unrounded variables, the numbers are exactly the same; however, Blizzard rounds everything to the nearest fourth significant digit.)
It has been noted across several expansions by several classes that their spells simply do not scale well enough with spellpower. If all spells scale by exactly the same value (since all coefficients are normalized), how is this possible? The short answer is talents. Although talents such as Wrath of Cenarius or Empowered Fire cause some of the scaling issues, they are rarely the true culprit. As they are the easiest example of this phenomenon, let's look at balance druids who use Wrath and Starfire. As mentioned, Wrath has a coefficient of 57.14% and Starfire has a coefficient of 100%. The Wrath of Cenarius talent boosts Wrath's coefficient to 67.14% and Starfire's to 120%. Using base cast times (2 seconds and 3.5 seconds, respectively), Wrath has an SPS of 33.57% and Starfire has an SPS of 34.28%. From this, it is easy to assume that Starfire has better spellpower scaling than Wrath.
There is, however, one other thing that impacts spellpower scaling: haste. Although haste is an entirely different topic that will be covered later, I will note that haste is an entirely normalized stat. No matter what the cast time of your spell, haste affects the scale of each and every spell in the game in exactly the same way. As with everything else in the game, there are a few exceptions to the rule. In these cases, the exception plays the part of talents that reduce base cast time. A vast majority of talent trees within the game have a talent within them that reduces the base cast of a spell by a flat value. In most cases, this value is .5 seconds (again, there are exceptions). In the case of balance druids, both Wrath and Starfire have their cast times reduced by .5 seconds via the Starlight Wrath talent. This is important because shaving .5 seconds off of a 2-second spell requires more haste than taking .5 seconds off of a 3.5-second spell. Since these talents alter base cast times of spells and therefore do not impact haste scaling in anyway, the only thing they alter is spellpower scaling. In the case of balance druids, the Starlight Wrath talent in conjunction with Wrath of Cenarius gives Wrath an SPS value of 44.76 and Starfire an SPS value of 40.
Due to the combination of different scaling values of talents such as Wrath of Cenarius and the impact on spell cast times from talents such as Starlight Wrath, it is entirely possible for a spell to have lower spellpower scaling than other spells, despite spellpower being equalized from a base level.
SPS is a theoretical value of scaling. While very useful for determining how spells scale with gear and as a tool for game balancing, it might not be exactly what you need. If you merely wish to see how much damage a spell is capable of doing at a given level of spellpower, then there is a simple formula to do that. The formula is mere Base Spell Damage + (Spellpower * Coefficient) = Spell Damage. Base Spell Damage is merely the damage a spell does as listed in the tooltip. For a spell such as Fireball, the base damage would be 888 to 1,132 damage. Since base damage for spells is usually a range of some form, using the range's mean (calculated as ((Highest Damage Point - Lowest Damage Point) / 2) + Lowest Damage Point) is the most accepted practice. Spellpower is whatever spellpower level you wish to use, and Coefficient is the coefficient of the spell in question.
Spellpower increasing rotational DPS
It is a fact of life (or rather, WoW) that players do not use only a single spell within their rotations. Instead, players often use multiple spells in order to achieve the highest level of DPS. Due to this, knowing the spellpower scaling of a specific spell is rather meaningless. There are some practical implications, sure, but the really important bit is knowing how spellpower affects a rotation as a whole. Here is the part where the theorycrafting takes a really sharp turn in becoming much more complex.
Configuring the effects of spellpower on a rotation is not something that can be done sitting at a desk with a pen and some paper. While the calculations themselves are certainly possible to do using this method, there is no way of getting the required data without actually going into the game. This is because calculating rotational DPS requires one additional piece of information: the overall damage breakdown of each ability.
Within a rotation, each spell used is going to account for a specific percentage of the total damage done. For example, a mage who casts three Fireballs that deal 100 damage each and then one Fire Blast that deals 100 damage would get 75% of his damage from Fireball and 25% of his damage from Fire Blast. Figuring out the actual damage breakdown for each class or spec in the game is impossible to do without logging into the game and finding out the data yourself. It also requires the use of an addon such as Recount or Skada, which will break down all of this information for you automatically in an easy-to-read format.
I should note that there is a method for getting this data without going into the game itself; however, doing so is either going to be very time-consuming or require the use of a simulator. In order to get this data without going into the game, you have to artificially construct a damage rotation by hand. This means writing out each ability cast within the rotation, running calculations to determine the damage caused by each spell, running calculations to determine the damage caused by the rotation as a whole, and then dividing each individual spell's damage by the total damage done within the rotation. Since a lot of rotations within the game rely on factors of RNG, any hand-written rotation is generally going to have to cover a time frame of at least five minutes of casting in order to be accurate. For some classes, this is going to add up to several hundred spell casts. That is a lot of additional math, and merely logging into the game to free cast a rotation at a target dummy is much, much simpler.
Once you have the damage breakdown of your specific rotation, you can then begin calculate the rotational scaling of spellpower. The formula for determining spellpower's effect on a rotation is simply (Spell A's SPS * Spell A's Damage Percentage) + (Spell B's SPS * Spell B's Damage Percentage), repeating the process for each spell involved. Again using a balance druid, say that 56% of his rotational damage is from Starfire and 44% of his rotational damage is from Wrath. Talented, Wrath has an SPS of 44.76, and Starfire has an SPS of 40. The formula is set up like this: (44.76 * .44) + (40 * .56), which results in a rotational SPS of 42.09. This means that a balance druid's rotation gains 42.09% of his spellpower value every second.






Reader Comments (Page 1 of 3)
raspybunk Apr 13th 2010 7:23PM
Splendid, never see wow.com cover this type of thing. glad to see it.
Terethall Apr 13th 2010 9:45PM
Indeed. I love articles like this, because this information tends to take a lot of time to search out if you try to find it yourself. As a raider, I like this sort of article because it helps me stay informed and because it makes those I raid with more likely to be more knowledgeable about their roles. As a gamer, I just like taking a look under the hood every now and then of a game I spend so much time playing.
The only thing this article was lacking was a mention of how other stats like haste and crit can affect spellpower's potency and rotational coefficients. I'm looking forward for the post on haste, and hopefully crit will get a nod. Also, I'm completely ok if you want to put off ArPen until Cataclysm if you know what I mean, Tyler. I don't think either of us would enjoy that article.
Rakah Apr 13th 2010 10:24PM
For anyone that can't be bothered with the maths part of the addon DrDamage allows you to see coefficiants as well as average dpc etc.
crown.lucas Apr 14th 2010 10:05AM
Glad, but kinda useless now TBH. Aren't they getting rid of spell power in Cataclysm? =/
Turtlehead Apr 14th 2010 8:43PM
"Glad, but kinda useless now TBH. Aren't they getting rid of spell power in Cataclysm"
Spell power is being taken off gear. No more +31 SP. We'll get it via INT on gear. The guts of the entire system which determines spell damage isn't being removed and replaced with one from the Hello Kitty MMO.
Defense, armor penetration? Also staying in the game once cut from gear.
Angus Apr 13th 2010 7:39PM
"Here is the part where the theorycrafting takes a really sharp turn in becoming much more complex. "
You know, I have read up on this stuff and you already had me cross-eyed at this point.
Great information.
celticmuse7 Apr 13th 2010 7:45PM
Thanks for the slap in the face example of why I'm not a theorycrafter.
/curl up in a ball
/cry softly
vertigobliss86 Apr 13th 2010 8:28PM
I'm with you on that one. As much as I enjoy the game (and keep it as a game) and enjoy understanding it better, all the equation stuff just averts my attention like a geek babbling about how he got this totally awesome blue item that totally boosted his stats on this totally interesting run for 20 minutes (never brought up WoW with him again lol).
To me, moar SP = better stuff. It is a game afterall, to each his own. Enjoy theorycrafters =P
PS: Leave a quicky at the end explaining what the textbook you just wrote said in a simplified form
ianordie Apr 14th 2010 5:56AM
I feel this way as well. As much as I want to follow what was being said in this article, my brain will not get past the fact it is a game I play to relax and should be treated as such.
That put aside, I do want to dip my fingers in this pie however it is all likely to change come cataclysm, so I may wait it out.
Still a very good article.
jaz Apr 13th 2010 7:47PM
This is great. But I'll be honest, I'm really ok with just knowing -
more spellpower=good
in other news, hot babes still like puppies.
raspybunk Apr 13th 2010 9:46PM
People will be more likely to understand the practice and the need if there is a theorycrafting 102.
In Cataclysm many of the damage increase talents will disappear, and at the same time we'll have more points to spend and nowhere further to go in our favorite trees.
I'm sure they'll cover damage per cast time and other such stuff soon, as far as when talents are calculated in the scheme of things and additive and multiplicative fun.
Anyways, with more "free" talents in our trees, theory crafting will be very fun. Especially for, say, rated BGs offer more in the way of "specialty" specs than arena. You can theory craft the highest dps talents in a disc/shadow build for more utility in the way of a silence, shorter fear CD, and your new slow, should you choose to go into shadow to that 21/22 talent point. Theorycrafting allows for fun thoughts, like how high can I keep my dps while becoming the best CC flag-runner guard on the planet. Or how high can I keep my heals...
Greg Apr 13th 2010 7:56PM
Wow- that is really complicated- er informative- yes, informative.
Please don't do an article on Hello Kitty Island Adventure 'sleeping power' coefficients and their scaling with wand upgrades.
I'll feel really dumb then. And probably cry.
More sleeping power = good. More spell power = good. All is right with the world. Of Warcraft.
Tyler Caraway Apr 13th 2010 8:10PM
Aww, but that was going to be the next bit. I guess I'll have to think of something else then. :(
On a more serious note; could you perhaps explain what part of the column was confusing? Was it too mathy, too wordy? Did I gloss over topics too much and not explain things properly?
Also, was the topic too broad? Instead of doing a generalized spellpower write-up, would it be easier to follow if it was more class specific? Spec specific? Spell specific?
I'm a guy that went from nearly failing basic Algebra in school to now spending my free time doing Calculus work for a video game. I went into theorycrafting knowing nothing, and with how spread out everything is, it was very difficult to just understand the mechanics. Reading up on my class was easy, but nothing really translated well into how the game works overall. I'd really like to show that to people.
I understand this sort of thing isn't for everyone, but I really do want to try and show people that theorycrafting in WoW is a lot more simple than it appears.
Cerbrusprime Apr 13th 2010 8:31PM
One thing that would really be helpful for scanning/reading the column is to separate the equations from the text. Put the equations in the middle of the page, with their own paragraph breaks, and that would aid readability.
Besides that, I appreciate the concise breakdown, and look forward to more articles of this type in the future.
Greg Apr 13th 2010 8:59PM
Wow- Mr. Caraway- I'm very pleased you responded to my comment.
Sorry for the semi-trolling there. I just wanted to make a joke- but a somewhat relevant one.
At a very quick glance- I think your column might be easily split into several columns. For example, there is probably enough material to simply write a short piece on spellpower scaling with damaging spells (direct casted, DOT, and channeled). A little more emphasis can be put on highlighting one applicable spell from each class to draw the audience in and extend the article. This would still serve to be informative and entertaining without being overly academic.
Other pieces could be written about spellpower and healing, bonus spellpower talents/items, history of spellpower (previous expansions), and the future of spellpower (Cataclysm).
Just ideas though.
Heilig Apr 14th 2010 12:08AM
"On a more serious note; could you perhaps explain what part of the column was confusing? Was it too mathy, too wordy? Did I gloss over topics too much and not explain things properly?"
The confusing part is the topic. There's no way to make it less confusing with text, it's just a complicated subject. Make some Visual aids next time. I didn't really understand why different cast times had the coefficients they did until i saw a bar graph of cast time/coefficient relation.
al Apr 13th 2010 8:04PM
awesome
gamerunknown Apr 13th 2010 8:07PM
Does anyone know if longer casted spells have a higher chance to crit? I know the GCD is a threshold that makes faster spells less desirable once you go past it, but other than talents, are there any reasons to use slower spells?
I'm a holy priest that throws in the odd bit of dps and I tend to DoT, but as I don't have any talents affecting dps spells as far as I'm aware, would smite spam be as much dps as using any other spell?
I assume since I'm below hit cap instants/1 sec cast spells are favoured since I don't lose as much dps casting.
Tyler Caraway Apr 13th 2010 8:18PM
That's a good question, but one that's harder to answer in a very generalized way.
Spells, for the most part, are very equal when looking at them without any talents to support them. Nothing within how the game works actually favors slower or faster casting spells. If you don't have any talents which effect your DPS spells, then the only thing that would change their DPS value would be base damage.
To try and answer your question more specifically. So far as I know, Holy Fire is more damage than Smite with the addition of the DoT. In fact, all of your DoTs should be your highest DPET (Damage per Effective Time - or the damage the spell does in relation to how much time you spend casting it) spells and you will want to use them. So, I think, it would be Shadow Word: Pain > Devouring Plague > Holy Fire > Mind Blast > Smite, but I've never actually done any theorycrafting for Holy Priest DPS since they aren't generally considered a DPS spec.
I'll try and ask Dawn about it next time I see her. Figuratively speaking of course.
Turtlehead Apr 14th 2010 8:46PM
I'd give holy nova a whirl. The group won't be able to claim you stopped healing!
Rereading your healing build with damage in mind you'll find a lot of dual use talents.
Someone (http://holyfirespec.blogspot.com/2009/05/smite-leveling-build-lvl-70-to-80-31.html) ran a blog for the rare smite DPS unicorns and his build highlights them. He also talks rotation and numbers. (It is a year old; talents may have changed since.)
Wasn't kidding about nova. Your lack of hit won't matter on heroic five trash. The overgeared tank is taking trivial damage, nova will cover it while DPSing w/o extra threat. Murders the mana pool but we're talking fun here, right?