Debunking the "Top" Mage Debate

[Akraen 230]

Didn't get a chance to run Arcane #'s, but I put this post on MMO-C because I'm fed up of people saying you need to be fire because it's "the best."

 

Both specs rock, all 3 specs rock. People need to understand how to interpret data and develop a keener eye on the difference between statistics (interpreted) and data (presented).

 

Source:http://www.mmo-champion.com/threads/1398132-Frost-vs-Fire-pve/page2#22

--

Statistics are created from data. The data may be pure, but the statistics are skewed.

 

If you want to compare the specs, compare the data. Do DPET & regression testing for the three specs.

 

You can't go by sims, because sims actually put frost ahead. Hell, look at this: http://www.wowprogress.com/simdps/us/stormrage/class.mage/

 

You honestly think that's accurate?

 

So how about raidbots, let's take a simple fight like Malkorok. I pick Malkorok because there won't be log-range issues, no special tasks, no alternate realms, no advantage to any spec, as any of the 3 specs can do the mechanics with minimal sacrifice to rotational function.

 

Arcane: a well timed Arc. Barr. would cleave onto adds nicely, could inflate #'s slightly with multidot

Fire: a well timed Combustion could inflate #'s slightly

Frost: a well timed Frozen Orb & icicle cleave could inflate #'s slightly

 

Pretty even advantages so let's begin some analysis:

 

Going to get the top 5 parses from US/EU and strike out the outliers. Why top 5? Well, I think your goal here is to know what's "best" right?

 

Frost:

1 Мишарики DPSe: 441519 Time: 04:53

Drjayfisto DPSe: 429776 Time: 4:47 - Received Tricks

2 Natubob DPSe: 423190 Time: 4:30

3 Adamas DPSe: 414621 Time: 4:40

4 Pete DPSe: 414621 Time: 5:02

Feylaa DPSe: 411877 Time: 5:28 - Received Tricks

5 Xiaozhii DPSe: 411531 Time: 4:49

 

Fire:

Jamiex DPSe: 558779 Time: 0:52 - Bogus parse

Runlikêhell DPSe: 480081 Time: 4:19 - Received Tricks

1 Karqon DPSe: 448802 Time: 4:32

2 Milks DPSe: 447131 Time: 4:23

3 Orby DPSe: 446829 Time: 4:11

4 Sfougz DPSe: 439730 Time: 4:24

Paybax DPSe: 430785 Time: 5:08 - Received Tricks

5 Dewgong DPSe: 429128 Time: 4:53

 

Fire mages are in better guilds statistically and appear to have better kill times. So let's calculate how much DPS(e) per second they did. So let's find the average Malkorok kill time for all the 10 frost and fire mages, and normalize each fight to that duration and see how it impacts DPSe.

 

Duration / AvgDuration = Norm.DPS / DPSe

 

So the new rank list becomes:

If each of them had a fight length of 4 minutes 37 seconds

 

Frost:

1. Мишарики 467022

2. Pete 452042

3. Xiaozhii 429359

4. Adamas 419112

5. Natubob 412496

 

Average DPS of top 5 frost mages: 436006

 

Fire:

1. Dewgong 453915

2. Karqon 440701

3. Milks 424532

4. Sfougz 419093

5. Orby 404888

 

Average DPS of top 5 fire mages: 428626

 

But if you look at raidbots, this is what you see:

 

Which is bad, because this is a much more accurate way of looking at it. And oh! Look! They're really close to each other and alternate Frost/Fire/Frost/Fire/etc

 

As I said before, old mindsets die hard.

Edited November 18, 2013 by Akraen

[Guest WeeWilly]

howdy,

 

read what was said on the other site, but since I am playing my first mage, I went with Frost as I read it is more forgiving.  Is that not true?   

[Akraen 230]

Not really one way or the other. Fire's pretty steady and more mobile. Frost doesn't have to worry about lining up a combustion. Both recover pretty well. It'll be personal, entirely personal, which one is best. You just have to try them, and don't forget arcane either.

[Oltier 471]

That is what I've been saying for a long time.

 

Finally someone approves me!

 

And this is the problem with raidbots. It is not checked so the "invalid" fights are also included, combined with the fact that most "top" mages choose fire.

 

Then, most players will follow the top players blindly and state that fire is better, in a stubborn way,  because THAT guy there used fire

[Guest WeeWilly]

oh, I think most folks approve of ya Oltier  :P  and those who don't can eat pickled herring. 

[Guest WeeWilly]

you mean I can move with "fire" instead of having to stand there and take it?  I really did not know that.  When I go home for the day, I will have to add fire as my 2nd one and try it out. 

 

One thing, I don't have the ability is day to day memory and learning a new spec will be rough.  But I think there is a fire mage guide here. 

 

Not sure about Arcane, I just need something that is forgiving for me.

 

As I don't raid or pvp, I really don't care about that aspect of the game (not meaning playing arcane, I just don't care about pvp or raiding).  Thanks Akraen.

Edited November 18, 2013 by WeeWilly

[Akraen 230]

If you don't care about pvp or raiding and you're just generally looking for something solid, consistent, easy to learn, no button bloat, and offers a lot of control...

 

 

...go frost :P

[nefigah 26]

Sorry for the stupidity, but I'm having trouble understanding the original post. I have three (somewhat unrelated) questions.

1) From what I understand, taking total fight length and dividing by your total damage done gives "DPS"; however "effective" DPS means something different. I seem to recall it has something to do with active time. (It would be nice if it meant "damage that was effective" as opposed to klaxxi-esque meter padding, but I know that's tough to determine.) What exactly is the DPSe calculation?

2) Now let's pretend I understand DPSe. You said:

"let's calculate how much DPS(e) per second they did"

So now we are looking at effective damage per second per second? That hurts my brain... Is that like looking at their "damage acceleration" as it were?

3) I don't understand the equation

Duration / AvgDuration = Norm.DPS / DPSe

The left hand side would tell us (I think) how much faster/slower a guild was at killing the boss compared to the average. But I don't get the rhs. Why would Pete the Frost Mage have a higher rank if his guild were able to kill Malkorok faster (assuming his performance didn't change)?

Thanks again and sorry for being bad at math!

[Storm 426]

I'm not super well versed with all details, but essentially DPSe takes into account things like Heroism/Bloodlust, potions, and cooldowns in general.

 

So, for example:

 

- in a 4 minute fight Hero/Bloodlust is up for 16.6% of the fight which roughly equates to 5% dps boost compared to having no hero/bloodlust at all. If the fight lasts 5 minutes then hero/blood was only active for 13.3% of the fight, which is only equal to a 4% dps boost.

 

That's just 1 major cooldown, and a 1% difference in your overall dps, just based on how quickly your raid group kills the boss.

[Akraen 230]

I'm more of a business-hardened math guy than an ivory tower/engineer.

 

Basically I am not good at math terminology. I understand statistic generation and analysis for accounting and financial structures. I'm a trend spotter, professionally I spot fake trends-- it's what I do for a living.

 

So what I try to put out there for people who play WoW is the encouragement of a skeptical mindset. Rather than defend the accepted conclusion, look at how people got there and doubt it. Doubt it to the core. When someone says spec Y is bad, don't believe them. Tear into the data like a mad man until you know something to be true.

 

Oftentimes I pick the wrong side entirely, that's fun. Like when I was trying to prove that NT is always best above 12684 haste. Well I ended up proving myself wrong, then I tried to prove it's best 13053, and so on. Thing is, my reluctance to believe it caused Lhivera and others to realize they weren't right either. It made one big gray area a bit clearer by now acknowledging it's a gray area and placing the 12684 to 15.8k haste range in a quarantined area on its own, where LB roughly equals NT.

 

I do it just to keep people thinking.

 

This is an example of that. I created an artificial metric, which is often used in statistical analysis to spot fake trends. Because if you can get a more logical trend out of rearranging parameters then it's a red flag for the original data trend. What the guys on MMOC don't understand is that median is the center of a data set, whereas average is the average of the data set. Both are horrible statistical tools for determining the performance of a spec. One is chopping off the top performers while the other is incorporating the bogus logs (52 second kill of Malkorok for 588k dps?!).

 

So what I did was attempt to extend the fight to a normalized duration, realizing that heroism and procs impact DPS a lot. The shorter the fight that is longer than 3 min 20 sec (b/c of cds), the higher the DPS. So between about 3:20 and 6:00, DPS gradually decreases.

 

Finding a common fight duration puts them all on a much more even playing field. So that's what I did. Then you can see it's pretty close. A real-world way to test this would be to simply have a few weeks of every heroic mage playing frost, then fire, then arcane. If they play the specs well, and try as hard as they can, we'd get better comparison data.

 

Until that point it's nearly impossible.

[Guest WeeWilly]

@ Akrean,

 

well after yesterdays irksome deaths from being fire and arcane - I learned fire and arcane are not forgiving and play a mage because it's something different than what I am use too, well, lets just say frost is where I am staying.  At least I tried it for a bunch of hours.  Thanks for an interesting post :D

[Elro 47]

I can't say which spec is "the best" anymore, but as someone who is pretty experienced with all 3, it comes down to playstyle, and in some cases, the encounter. Frost has better burst aoe than Fire, but that doesn't make it a better or worse spec, just different :) I find myself spec swapping every couple months on my mage, just because I get bored, and like to keep it interesting. Part of that probably comes from Cata, where it was Arcane in FL, and Fire in DS, and that was it. Ah well :D

[Belicosa 11]

The best spec for you is the one you can play best. For me that is Arcane. Precise rotations make me more comfortable (which is why my lock and rogue are lagging behind), And I can always count on some nice burst crits to drive me steadily up the dps chart.

 

I will say that frost is very good for levelling though, as your survivability is increased (although rune of slow kind of levels that out). Now to figure out if I can boost my mastery to 100 without sacrificing too much haste before the expansion...

 

http://us.battle.net/wow/en/character/twisting-nether/Belicosa/advanced

 

Please let me know if you see any glaring holes that I have missed (FM disorder makes me fuzzy sometimes and I make stupid mistakes, like not having glyph of vendetta on my rogue for assassination spec).

Edited November 28, 2013 by Belicosa