Tuesday, July 20, 2010

DPS Scaling in ICC: Another Angle

Curiosity struck again as I was wandering around in World of Logs, so I thought I'd do another post on dps scaling. My previous posts on the subject can be found here, here, and here.

One of the comments (among many good ones) was that looking at only the top 10 of a class didn't get at what was the 'typical' dps in a raid situation. To which I replied: I don't care about 'typical' dps, I'm interested in what can be accomplished under ideal conditions. I'm still not interested in 'typical' dps. I want to know what is the best dps each spec can do and how each class/spec has fared in regards to max potential as we gear up in Icecrown Citadel.

Another angle at the top end of dps performance is to look not at the average dps of the top 10 of each class, but to simply look at the top 100 dps overall, and see what the class composition for each boss fight in ICC is. If 50 of the top 100 dps on a particular fight are, say, arcane mages, then that suggests a couple things: 1) arcane mages do good dps on that boss (relative to other specs), and 2) that particular fight is very arcane mage friendly. If arcane mages are very well-represented in the top 100 for all the bossfights, that suggests that mages do very well overall compared to other specs, and that they have scaled well in current content.
My numbers aren't perfect, because there were a few scattered arms warriors that hit the top 100 here and there, and a couple of survival hunters or destruction warlocks on a fight or two. There were maybe a total of 7-8 of these specs combined, so when one of them was in the top 100, I just moved to the next dps below them to include in the chart.
Something else I glossed over last time was the sample size. In general, this 'sample' is fairly small - only 100 players for each boss, for a total of 1100 players on each difficulty. However, the number of potential players ranked - that is, how many rogues/shaman/druids/etc are parsed on WOL - is in the tens of thousands. There are typically 125 pages (40 each page) of ranked players for each bossfight. So, while my sample is not representative of what each spec 'typically' does, I think it's fairly safe to say my sample is representative of the 'best' each spec has to offer.

The number crunching for this post was, thankfully, far easier than for the last one. All I did was count how many of each dps spec was in the top 100 dps for each bossfight. The results are below:

As one would expect from this slightly different take on the same question, most of the same basic conclusions of my previous scaling examination remain the same. Still, take a look and see what your thoughts might be with this variation on the theme.

Looking at the 25 man normal difficulty, a few things jump out at me.

First, of course, is that fire and fury are just as dominant as the previous post suggested. 43% of the top 100 dps over all of the ICC bossfights was one of these two specs.
Combat rogues, in third place, didn't even make up 10% of top dps, and it went downhill from there.
Two specs, 43% of the top dps. The other 14 'viable' dps specs? 57%. Ouch.

And that's only on normal difficulty. Let's take a look at the heroic chart, where scaling really starts to take hold.
1) Fury is back on top by an incredible 2-to-1 margin over its nearest competitor.
2) Fury and fire now account for almost 67% of top dps for all of Icecrown, leaving a mere 33% for the remaining 14 specs. Um, something is not balanced in the State of Denmark.
3) On heroic Marrowgar, a whopping 79 fury warriors were in the top 100 dps. 11 specs didn't have a single player ranked. That means that the 79th-best warrior did more dps than the best hunter, warlock, DK, shammy, priest, .... well, you get the picture.
4) Overall, fury has more players in the top 100 than all 14 non-fire specs, combined!
5) Interestingly, it appears that unholy DKs and shadow priests did fairly well - as opposed to their showing on the previous examination. It's clear, though, that this is entirely due to the fact that both specs have 'specialty fights' - fights where they seem uniquely suited to a good performance (spriests on Valithria, DKs on Sindragosa and LK). Take away their specialty fights, and they combine for about 3% of top scores.
6) Six specs couldn't even manage an average of one score in the top 100, and poor enhancement shammies didn't have a single top-100 score in all of ICC.

EW. Dead horse. Beaten.
I'll continue to do the best I can  in my role as a dpser, and focus on competing with my spec peers, rather than bashing my head against the wall of superior warrior scaling.
That's all the have nots - everyone but fury and fire, really - can do.
And keep our fingers crossed for Cataclysm.