Frost mage: effort vs reward

[visc]

Mod edit: Changed title to be more informative

Frost Mage skill is negatively skewed. To a gross extent when compared to every other single spec in game.

This is the overwhelming problem with frost. No one wants to play, because the variance in skill is negatively skewed.
Celeste talks about how they have to be careful balancing damage because of skill differences in player groups (i.e between fire and frost), which is a genuine problem. https://twitter.com/Celestalon/status/8 ... 7928570881" target="_blank

However, the underlying cause of the problem is that the variance in the skill required of frost is too high. AT THE LOWER END.





Here is a bell curve of how frost skill affects dps. We can see the skew of to the left. The player is harshly punished for playing at the mean skill level.



Here is a fire distribution over the top in red. Here we can see the mean and average player can perform better at a lower skill level.

These distributions force players players who care about their dps and are below the top 5% of skill to avoid frost.
There is no reward for playing significantly above the mean in frost to entice players. There is no reason to play a spec with a distribution like this.

Why play frost if they can play fire and have a slightly positively skewed skill cap? If this can't be balanced, if there is going to be a punishment for performing at the mean, there needs to be a reward for performing far above it.
Or else players will simply play fire. As fire, there is a similar reward for playing far above the mean skill plus a bonus when they fall to the mean or below.


Don't balance the dps. Balance skill variance (or increase rewards for increased skill variance), then balance the dps.

Here some Wow logs that show the skewed distribution across lower to mid skill levels.

10% normal NH


50% normal NH


Unless the plan is to have one scapegoat spec always at the bottom that no one plays so there are less complaints. Which I'm beginning to think may be the case...

[Muphrid]

Analysis of this sort that doesn't discuss whether the players of each spec actually have the same skill distribution is meaningless. How do you know if a fire mage at 75th percentile is equally skilled to a Frost mage at 75th percentile? Or how different they are?

[visc]

Analysis of this sort that doesn't discuss whether the players of each spec actually have the same skill distribution is meaningless. How do you know if a fire mage at 75th percentile is equally skilled to a Frost mage at 75th percentile? Or how different they are?

I don't know if fire and frost mages are equally skilled at the 75th percentile.

I linked the 10th percentile, and I think it's a pretty safe bet that fire and frost mages at the 10th percentile are equally skilled.
I think it's almost as safe at the 50th percentile. They are the logs I linked, not the 75th or 90th percentile. And those lower percentiles are the ones relevant to what I discuss, not the 75th or higher percentiles, where skill may significantly differ.

[Pouponu]

So frost sucks because it's too easy or are you saying the other specs are too hard?

[Xzaar]

He's saying the penalty for not being very good at frost is too high compared to the penalty of not being very good in fire.

Frost is actually performing quite well if you play well, but it should be way higher considering how penalizing it is to be subpar at it.

[glasscannon]

He's saying the penalty for not being very good at frost is too high compared to the penalty of not being very good in fire.

Frost is actually performing quite well if you play well, but it should be way higher considering how penalizing it is to be subpar at it.

Fantastic point. Which inherently brings the balance conundrum. I like to think I'm performing very well as frost in my 25 man guild raid, so I see the overall NH class stats and scratch my head. The problem being, if they buff to the mean so that lower skill players are more in line, what happens to those who are already mechanically sound at frost? Do they get subjected to a mass influx of forum posts demanding nerfs to 'OP Frost' all of a sudden ?

[Pouponu]

He's saying the penalty for not being very good at frost is too high compared to the penalty of not being very good in fire.

Frost is actually performing quite well if you play well, but it should be way higher considering how penalizing it is to be subpar at it.

Ok, yeah, I was confused there. I'm an average skilled mage. I understand the aspects of all of the specs, but I make my fair share of mistakes.

[Muphrid]

I don't know if fire and frost mages are equally skilled at the 75th percentile.

I linked the 10th percentile, and I think it's a pretty safe bet that fire and frost mages at the 10th percentile are equally skilled.
I think it's almost as safe at the 50th percentile. They are the logs I linked, not the 75th or 90th percentile. And those lower percentiles are the ones relevant to what I discuss, not the 75th or higher percentiles, where skill may significantly differ.

I don't think you can know whether they're equally skilled at any percentile because there is no way to prove it one way or the other.

You simply cannot use any Nth percentile across specs to try to infer performance as a function of skill. The distribution of skill across the population of mages playing each spec is totally and utterly unknown.

[Cnboxer]

Nice job Visc saved me from writing about it.
I was quite surprised when I saw even 95th percentile Frost Mages were dead last in the rankings, as my logs never seemed below par to other classes/specs in raids. Wouldn't say Frost is a top tier spec but it is probably just above middle of the pack and still excels on extended ST fights, 2 target cleave etc. We are definitely better now than before patch.
It is hard to believe that only 5% of Frost Mages are able to play the spec competitively and that is something that should definitely be addressed. Preferably it's not made tooooooo though as the difficulty of the spec is also what makes it enjoyable. Not because no one else can do it but the constant drive to improve more.

[Padde]

Definitely agree with your point. As I see it there are 2 reasons to play frost, either you enjoy the playstyle or you have the legendaries for it. increasing our dmg via aura buffs simply shifts the distribution that you presented to the right but the shape of the curve is the reason behind the frustration. Thanks a lot for this post .

[Mage]

I don't think many people frequenting this forum care what things look like at 10 or 50 percentile.

[Hidden]

10th percentile is pretty much the "You died/DCed." percentile so I'm not sure why you'd ever take that for anything. Looking at e.g. Krosus, 10th percentile has Havoc as a distant last because you can charge down the platform and die.

In general, percentile rankings are affected by much more than just skill. 10th percentile, for example, would be expected to be a mix of low skill, low information, low gear, lack of traits, lack of legendaries, deaths and bad RNG. By low information I mean players which are uninformed about their spec - that's likely a larger part of the players playing non-meta specs such as Frost.

If we took your own metric for performance at different skill levels, by the way, your charts would be incorrect. Frost underperforms by roughly the same margin compared to fire at any percentile >=50%.

In general, it seems very disingenuous to post these plots, implying their accuracy, when they're completely unsubstantiated for the following reasons:
1. Percentiles don't directly relate to skill (see above).
2. You cannot reasonably generate a plot from no more than two data points.
3. The plots don't even match the available data points.

Lastly I want to add that your title is also complete clickbait. Frost performs worse than Arcane/Fire at the very highest level as well. That's the level most players look at to decide which spec to play - no matter if they don't play at that level themselves.

[kobeathris]

I think a bigger problem with Frost, and TV specifically, is the degree to which it is impacted by RNG.

Suppose you have a fight long enough to get 2 uses of Icy Veins in. Over the course of that fight, it's probably long enough to get fairly close to the expected number of brain freeze and fingers of frost procs, but because of how TV works, going on a run of 20 seconds without a proc followed by 8 in a row, vs 8 in a row followed by 20 seconds without will be a massive swing in dps even though the total number of ice lances is the same. If they changed Thermal Void to something like "Your Ice Lances against frozen targets reduce the cooldown of Icy Veins by .5 seconds", the TV wouldn't be hit with the double randomness of being both reliant on procs, and having to have them all hit within a certain window to be useful.

[Drozzy]

Frost performs worse than Arcane/Fire at the very highest level as well. That's the level most players look at to decide which spec to play - no matter if they don't play at that level themselves.

This is a 2-way street. I would say that frost actually outperforms both fire and arcane (In single target and cleave) at the very, very highest percentiles. The problem largely arises from the lack of frost mages, in addition to the lack of very skilled frost mages. For a quick example, The highest public frost mage parse on mythic Skorpyron (with the 1 day of parses we have, on an aoe fight, where frost is considered a 'weak' spec) is Thingy with 771k dps, which is ~#200 parse counting all public logged mages. Thingy is an outstanding player. Having the same legendaries as Thingy, in my own mythic Skorpyron attempts last night (we did not down him, so there are no rankings), my pulls where I didn't misplay and no extra skorpids were pulled, I was pulling 990k dps. Which would be ~#80 rank. And I only have 40 traits in my weapon. Frost performs very well when played perfectly, but even the best of the best can't play frost perfectly without a lot of practice on a boss. There are 5 public logs of frost mages on mythic Skorpyron. There are about 240 fire mages.

[Poosh]

I think frost is better than both specs on most fights, but you wont catch me playing it on skorp. I literally have 35 points in my fire weapon just for the AOE fights like that. I have been top dps in my guild doing heroic progression this week when I play frost.

[Ulfin]

From my recent experience on the new raid with frost using thermal void, there is an enormous gap on the very few pulls I can extend icyveins over 1 minute and benefit from a virtually unlimited duration of the buff, getting back the cds to extend it, and the numerous unlucky attempts when I don't get enough procs of brain freeze and finger of frost and fall short for 6 to 8 s.

This being said :
- the T19 2 part bonus set brings you extra brain freeze therefore extra chances to extend it
- the T19 4 part bonus set brings you an extra finger of frost every minute or at least an increased chance extra procs after the first automatic finger of frost you get on first tic/target
- Lady Vash's grap gives you an extra finger of frost every 10 s
- some relics reduce your icyvein cd and should ease this exercise

All this elements tend to prove with proper gear and execution it might be possible to extend icyveins duration to 100% of the fight bringing important improvement of dps.

As was noticed earlier by a member, most mages started with fire and are not yet ready to use frost to it's best, although frost single target dps is very impressive with perfect execution in NH. The few that forsaw that are ahead by a large margin. If that ability to anticipate the new situation is what you call a punishing difference of skill then yes indeed, lack of anticipation made a large number of mages underperform with frost this days, yet I'm sure the gap will narrow in the next few weeks, regardless of the eventual changes of inner mechanism of frost magic.

[Xei]

Let's really break it down to the reasons.

With NH being cleave heavy, and many fights being 4+ targets cleave heavy, fire is always going to do more "DPS". Uninformed raid leaders (which, let's be honest, would likely be 90%+ of them) will tell Mages to spec fire because it does "more DPS". But will the better cleave Fire brings actually help you kill bosses better/faster?

How many times has your raid wiped on Skorp because the little crappy adds didn't die in time? Probably never.

How many times have you wiped because the priority DPS add didn't die in time? Probably quite a few times - or at the very least caused a death or burnt some defensive CDs or exhausted healer mana.

Frost is GREAT at priority target cleave. It will nuke down the priority target MUCH better than fire. Ice Lance will crit hard on that add, and do 85% damage to one next to it - with Blizzard ticking and FO etc etc. Fire will likely just be LB and DB with some ignite/conflag DoTs.

I couldn't be bothered checking through logs, but if I was to find one of the fights where there is priority DPS on adds and it is critical to the outcome of the fight that it dies quick and check which spec did more damage to that add (excluding CDs as that will skew results) I would has a guess that Frost would be superior.

That is why I choose to play Frost. If that wasn't the case, and we couldn't beat a boss because we didn't have enough incidental cleave slowly ticking on 4-5+ mobs, then I will be the first to respec Fire and use that. But it hasn't happened yet, and I doubt it will.

[Mage]

Frost boss damage is pretty much the same as other specs, with aoe lacking https://www.warcraftlogs.com/statistics ... ic=bossdps" target="_blank

AKA - rock bottom.

[pleblius]

I'm not a bad Frost mage, so it's difficult for me to add personal input to what you're saying. However, one thing I do notice is that I think, overall, most Frost mages aren't very good.

I've stepped down from raiding for awhile. Personal reasons. But I did a Normal NH with my guild last night. I was ranking 95+ percentile on almost every fight, and in the top half of DPS in the raid. I was also beating two of the three Fire Mages on every fight.
I had Sephuz and Prydaz, 895 equipped ilvl with a 900 Icon of Rot and a 860 Construct Core as my trinkets. No two-piece.
I don't think I should be ranking that high. My Sim DPS was like 500k ST. After I got 2-piece and a weapon upgrade last night, my sim DPS shot up to 600k. My ilvl is barely above M EN level. I only have 46 traits in my weapon. I have zero throughput legendaries. Etc.

I don't really know how to put it, but when I got in an argument with somebody in my guild about how good Frost Mage was, and he brought up log distributions, all I could say was, "Eh, I've been really competitive in it. All I can guess is that most people playing Frost are really casual."

I'm curious about this phenomenon, though.

[Seprest]

I improve a little attempt to attempt, but I am getting 40-60 percentile. Not sure if dps padding is that rampant on some fights, but with 888ilvl I was only be to pull a little over 400k dps on Krosus, with light movement as I would always double blink to the back for add pools....but still that only landed me in mid 40s percentile for my ilvl. No 2set, I am at the shatter cap ~33%. Meanwhile 2 rogues pulling 500k+ with low 880ilvl gear. Pugging throughout these last two weeks I could compete with arcane and frost mages of similar ilvls, but going 6/10H I definitely feel like there is some secret to frost that has been hidden with the patch changes and secondary reductions. I sim haste as having over twice the power of crit after shatter cap.

In EN I was able to consistently get 70-80 percentile, often getting my next icy veins before the first ran out on many fights in NH Heroic. As a casual raider, but one who has played since vanilla, I felt like I could easily be in the top quarter of frost mages, but not any more. Something has changed, or perhaps me and other frost mages are not playing CORRECTLY any more after the patch.

*Posting to give a casual raiders perspective. Don't feel gimped compared to most mages I run into, but compared to some classes, and apparently the mages on warcraftlogs, I am.