Get the Desktop App for Battle.net Now
- All your games in 1 place
- Log in once
- Automatic game updates
I posted this on MMO-Champion for additional discussion. I'm hoping for this to get to developers:
I'm more and more concerned with each passing parse that Frost isn't scaling well enough. Echoes of Dragon Soul all over again. Frost really was viable in Tier 12, but then scaling killed it in Tier 13.
I don't think this is necessarily a frost vs. fire thread, more of a frost vs. frost thread.
I don't claim to be the best player in the world, but probably top 1%. I'm pushing the limits of frost as best I can in every fight right now (much to my guild's chagrin), and I'm finding on every fight I would be better off fire. This is concerning me. In the first week of Mogu'shan when people were wearing a lot of blues, my frost numbers were exceeding most fire mages on World of Logs.
So let's look at aggregate numbers over time (data from World of Logs):
All DPS numbers are median DPS scores from all players of the spec listed submitted to WoL.
Stone Guard 25H
October 25th: Frost DPS @ 127198.5/ Fire DPS @ 186551
November 4th: Frost DPS @ 149517 / Fire DPS @ 218040.5
November 12th: Frost DPS @ 144261 / Fire DPS @ 238016.5
The difference between Frost & Fire at the start was 59352.5, progressed to 68523.5, and most recently the gap has widened to 93755.5 dps difference between the two. Frost is beginning to look less viable and more like Dragon Soul numbers, but this extreme gap is likely due to DoT spreading.
Feng the Accursed 25H
October 25th: Frost DPS @ 77436 / Fire DPS @ 88450
November 4th: Frost DPS @ 77365 / Fire DPS @ 93675.5
November 12th: Frost DPS @ 80662 / Fire DPS @ 103494
Difference for this single-target (mostly, small adds are up briefly in one of the phases), starts at just a 11014 DPS difference on October 25th, progresses to a 16310.5 gap, and now more recently is at a 22832 DPS gap, with a major advantage to fire, on a completely different style of fight. Fire mechanics allow for DPS uptime to remain higher during epicenters, but overall despite what simcraft once said, there is no evidence to suggest frost dominance in single target.
And it's getting worse. Bugs don't like fire:
Imperial Vizier Zor'lok 25N
November 5th: Frost DPS @ 58450 / Fire DPS @ 80305
November 12th: Frost DPS @ 66383.5 / Fire DPS @ 88217
Gap goes from 21855 on the first week, and decreases to 21833.5. This fight remains consistent, though a 22k dps gap between specs is rather high, no? Remember, this is a median. Unless there are tons of really horrid frost mages dragging the rest of us down, that's quite a gap. To investigate this fight further I'll use my own numbers:
I'm able to push out World #29: 67326 dps. That's a bit higher than the median. Modoshi is the curve killer, he managed to do 96096 dps, score one for frost! However, our hard work isn't paying off, top fire DPS is Cimsta @ 106597, over 10k higher than the best frost mage. Thing is, Modoshi's DPS, if it were placed on the fire mage chart, would be down at #13. Not bad you say? Well, Modoshi is a one-off. Next highest frost mage is Eliron at 83231, still a lot higher than my own parse of 67326. If Eliron and I were on the fire chart, he would be #90, and I would be cutting myself at #221, I've never been that low. So there are many more fire mages able to consistently be much higher than the top fire mages.
November 5th: Frost DPS @ 71974 / Fire DPS @ 108095
November 12th: Frost DPS @ 70367 / Fire DPS @ 123880
Gap went from 36121, fire > frost all the way up to 53513. An 18k dps increase over one week across all median DPS! Fire is pulling ahead at an incredible rate.
One more, then I have to get back to work:
Amber-Shaper Un'sok 25N
November 5th: Frost DPS @ 119384 / Fire DPS @ 160991
November 12th: Frost DPS @ 133332 / Fire DPS @ 194731.5
Gap went from 41607, fire > frost all the way up to 61399.5.
In conclusion, for now, I think that these numbers are astounding. This gap is unbelievable. I do not think frost is too low because when I compare myself to other DPS classes, I feel pretty good. Then when I analyze my numbers versus fire-- I just feel like as time goes on it's going to be moving further and further back into Dragon Soul misery, where if you're not throwing fireballs, reroll.
I'm really pushing to be the frost mage this tier, the one that proves it is possible. But from the looks of it, I'm really afraid as I get into Heroic 25M HoF & Terrace, frost is going to fall apart entirely. I really would like it if this data can be commented on by blues, or that the MMO/WoW community can add more information to it. The pressure is really building. Mage will soon just be one spec. :(
Doesn't every pure dps class have 1 spec that's best in pve? Leave frost alone... making it so boring in pvp.
Edited by Smoovie on 11/13/2012 10:40 AM PST
I think frost is fairly solid single target relative to other specs, but that it does fall behind on cleave / aoe fights. I agree it could use a boost, but something to keep in mind is that because fire does higher theoretical damage in raids, high end raiders will spec into fire thus improving the median/mean relative to frost.
Maybe this is a lazy suggestion, but I think just buffing Frozen Orb to non-player mobs would be a great way to boost sustained and aoe damage. I don't think frost needs to be better at aoe than fire, but it needs to be competitive because as you have noted, raiders are overwhelmingly fire right now.
Please show me the numbers to back up your thoughts. I encourage opinions. Trust me, I am the biggest defender of frost (my guild wants to kill me).
Also, I chose 2 very easy 25M heroic bosses and 3 normal modes. This is where my progression is, I'm 492 ilvl, and I'm frost. So I don't think you are seeing my point. The scaling of frost v. fire, and the fact that you need to be one spec in order to progress.
It looks like you're using Top 100 as your sample size. Looking at the median for all parses, the difference between fire and frost on Stone Guard H is 17k.
Now, if you change that to 90th percentile, that difference becomes 2k.
The reason for this, is because there is very little frost data. Raidbots has 101 frost parses for Stone Guard 25H vs 2500+ fire. That means using Top 100 will show the highest outliers of fire against almost the entire data set of frost.
The same thing with Feng: http://raidbots.com/dpsbot/Feng_the_Accursed/25H/all/14/60/p90/#3
90th percentile parses show fire and frost to be less than 1k apart.
Edited by Nyteshadè on 11/13/2012 11:06 AM PST
I don't think the median of the parses above the 90th percentile is a raw enough sample. When you take 100 you're allowing 100 good, bad, and otherwise parses to demonstrate the class' capability. The sample size is simply larger.
You bring up a point about the outliers, but that point is mitigated by the point itself-- no parses = nobody is playing it = time to change something.
Nah - something is not right with raidbots when you measure by percentile.
Look at Gara'jal: http://raidbots.com/dpsbot/Gara%27jal_the_Spiritbinder/25H/all/14/60/p90/#3
There's no way that's accurate. 168 fire mages are above 100k, with a max out of 123k.
There's one frost mage above 100k. One.
Except with the current sample sizes, you are taking 100 good, bad, and outright terrible frost samples and pitting them against the 100 best and luckiest fire parses. With 90th percentile, you are taking the 10 best frost against the 250 best fire and coming in neck and neck.
However, that has always been the problem with mage specs. Any difference in DPS, no matter how small, becomes the default raid spec every tier. It also doesn't help that fire tends to have better utility through dot spreading which makes it heavily favored on many fights.
Nah - something is not right with raidbots when you measure by percentile.
So what's not accurate about that? There's 168 above 100k but if you keep going down, it falls sharply. The last ranked parse at 217 is only 71k. For 90th percentile of all parse, there would be another 25 or so that aren't displayed on WoL rankings to account for.
The only thing I see wrong is that it looks like raidbots may actually be displaying averages instead of median. IIRC, median is supposed to be the line where half of your data is above it and half below.
Because when you have so very few parses as frost, the 90th percentile includes the herpderps that died halfway through or aren't good, and lost mega dps.
90th percentile for fire is still going to include a higher better performing sample, because most of the better players are playing fire because they're required to for progression.
I'm one of a small handful of frost mages out there that gets to do this content because I'm the GM and do what I want :P
Look at all parses, probably the only way to really take the bad with the good when there's such a disparity among number of parses from each spec:
and so on...
Actually, that's the other way around. Let's look back at Stone Guard Heroic 25 again:
There are 101 frost parses on Raidbots vs. 2547 fire. Top 100 means you are using 99% of the entire frost data set (including the worst of the herpderps) and putting it against the best and luckiest fire parses. That's why your original numbers are so skewed.
90th percentile means you're looking at the top 10% per spec. That means for Stone Guard, you see the top 10 frost parses pitted against the top 255 fire. None of those players will have died (or at worst, died at the same time the boss did), none of the will have played poorly, and there's enough fire parses that it's not just the ones that lucked into the best crit chains and subsequent combustions.
In which case, it's hard to complain about the numbers themselves so much as people's perception of what they may (or may not) be actually capable of.
As I said earlier, mages have always been really bad about always favoring one spec no matter how close the DPS difference is. There have been times when fire and arcane were within 3% of each other and there would be rampant hate on those specced for the lower one.
And again, it doesn't help that many current raid mechanics favor dot spreading and cleaving. Glyph of ice lance helps a little bit but it doesn't even remotely compare to spreading ignite and combustion. Frozen orb is helpful but the cooldown is prohibitive except in cases like Elegon once you've gotten the adds grouped back up on the boss.
I am worried about frost's scaling though. I know it can't be used as much of an indication, but seeing how frost and fire still had a HUGE gap in heroic DS post 5.0 changes, it makes me wonder how well frost will scale in later tiers. Especially when I look at Simcraft and see higher stat values on everything for fire.
Indeed, more than any current gap, I'm worried about scale factors. Some things that could be done: Frozen Orb CD scales with haste, increase the Shatter multiplier (while lowering the base crit), make mastery work with Frostbolt for raiding in some way. The poor scaling compared to fire is a point of serious concern for the future viability of the spec.
I'm confused what you're arguin for here. If your argument is that the top 10% can still include herpderps, doesn't that just make your argument of usin top 100 even worse?
The answer is that yes, yes it does.
Blizz also buffed frost armor right before the expac, and the water jet ability for our pet is clearly designed to help out single target raid dps. I also thought I read that bloodlust/timewarp/heroism is giving us +spell haste instead of +casting speed which would benefit frost more but I'm not sure if that stuck. The point is they clearly want it to be a viable dps spec, and as the more complex of the two rotations, I'd be disappointed if it was doomed to forever be behind fire.
Threats of violence. We take these seriously and will alert the proper authorities.
Harassing or discriminatory language. This will not be tolerated.