Akraen's (in)Complete Frost Stat Analysis

[lynx 19]

Negative sir, he fixed the mastery calcs.

 

And please, go into more detail. I speak with Lhivera regularly and I'm sure he'd like to know your wonderous insights.

 

I gotta take a step to the side here and air a little frustration. Not directly at you lynx, just in general. How come so many people will critique the work of a very few people, but not contribute to finding workable solutions? When I first began trying to dissect frost mage in Tier 14 and posting findings on forums, I said flat out (and have many times since) that I am by no means a math guy.

 

I'm a wordsmith, a creative mind for intense pattern recognition and conceptual analysis. Even professionally, my job is to digest an incredibly complicated software system and relate it to businessmen in sensible terms. I love to read and study everything from philosophy to quantum mechanics. But math? I just never put the time in and my degree didn't require much more than a couple calculus courses.

 

I'd absolutely kill for you sleeper experts out there to actually help me out. It just does me no good to point out flaws and not offer solutions. I'd rather you simply not exist or not post in your current form. If I ever quit theorycrafting and the mage community it'll be because of the type of burnout you experience when you spend hours and hours obsessing over solving a problem only to find one know-it-all on the forums tell you that you're wrong.

 

All I want are answers. I don't care what the answers are. And sometimes it feels like I'm the only one trying to find them.

 

Mastery still seems broken (multiplies by 2). Are we talking about the same TC-Lite?

http://lhiveras-library.com/wow-spells

 

Dots... here it goes.

Lets start by how Lhivera does it. Assume 0% crit.

NT damage: (3899 + 2.928*sp)*Floor(12000*SpellHaste/1000 + 0.5)/12

There are few errors here.

3899 and 2.928 are incorrect values, it should be 3900 and 2.9232. But this is minor.

Floor(12000*SpellHaste/1000 + 0.5) this is wrong though as number of ticks isn't computed this way.

Right way to compute the number of ticks is this:

B=Base duration of the dot (in ms)

I=Base interval between ticks (in ms)

H=Spell haste (final value in char sheet)

Ticks = Round(B/Floor(I/H+0.5))

Where Round is banker's rounding.

This is what AMR tables have. It's in SimC as well. And most importantly, tested by yours truly, it's in the actual game. Since DPET is derived directly from number of ticks, this is where the difference comes from.

It isn't big, but it's crucial if you want to be pinpoint precise on your desired haste.

Edited April 8, 2014 by lynx

[Akraen 230]

@lynx:

If you look at the values I have input in my screenshots you'll see the mastery value is divided by 2. That's not "broken" as much as, hey keep in mind to divide mastery by 2 when using TC-Lite, lol.

 

Thank you for that clarification, helps verify what I thought. 18960 it is, forget that extra 9, lol. Thank you, it's easier for me to work with AMR anyway, but the robot never shares his math, even with people who want to help the robot...

 

@muphrid:

Good call on terminology, I'll change it to "previous DPET", it doesn't really change anything-- just how it's presented and I agree is easier to communicate.

 

It's the boss-to-boss fluctuations in ET%s that make it a lost effort. Everything from strategy to mechanics to 'xyz died go rescue their side on malkorok' completely mess up ET%s. I've always found in hectic, unstable raid environments, there's an edge to haste, but not one I could ever begin to quantify.

[Akraen 230]

Received clarification, it is 18960, AMR is correct. /phew

 

I have an update in the works to finish this off and take a break while I finish getting 14/14H and do beta testing. Conclusion is the same but by a lesser amount. Working in a few ET projections to make some happy, doesn't change the answer though :)

 

Kver reminded me of my own thread about the 19347 meta or time warp living bomb/NT cap (what would normally be the 28k) - that gives you sufficient damage gain to go ahead and jump up to 19347, but you really don't want to get much more haste than that unless we end up in ilvl 608 gear or something in a pre-6.0 Mythic SoO. If that happens, sod it I'll just aim for the highest number of haste before the squish deflates the epeen.

[Kver 65]

I didn't even know you made that thread, I just saw you and Frosty argue over breakpoints and I thought you missed the biggest one(In my opinion).

 

For just 400 extra haste you gain another extra LB tick with 30% haste(Either meta or BL), both of which you should have plenty uptime on. Even on Garrosh HC progression(10 minute fights) I get 30% uptime on meta, which means that you're paying a really low price for an extra LB tick quite often, considering you're supposed to snapshot that properly.

[Akraen 230]

Yeah that's how my guide got started-- I listed all the breakpoints and what they all meant. Then when I saw 14242, I more or less stopped thinking about other breakpoints pretty much until now.

[Kver 65]

Yeah that's how my guide got started-- I listed all the breakpoints and what they all meant. Then when I saw 14242, I more or less stopped thinking about other breakpoints pretty much until now.

And this, ladies and gentlemen, this is how theorycrafting works. You don't need any mathematical knowledge, a PhD in physics or anything of the like, you just look for a pattern, trying to think about how something works. That, literally, is all.

[Frostedmages 3]

I didn't even know you made that thread, I just saw you and Frosty argue over breakpoints and I thought you missed the biggest one(In my opinion).

 

For just 400 extra haste you gain another extra LB tick with 30% haste(Either meta or BL), both of which you should have plenty uptime on. Even on Garrosh HC progression(10 minute fights) I get 30% uptime on meta, which means that you're paying a really low price for an extra LB tick quite often, considering you're supposed to snapshot that properly.

 

The up-time of the meta for Frost mages is one of the highest (I think only  Boomkins and maybe Ele shamans have higher?), if you're lucky you can get upwards of 40%. Good idea.

[NickSeng 0]

FWIW, it's not that Lhiv's TCM is wrong, it's that people not inputing the values of mastery correctly. The value of Mastery in the tool is actually the rating, not the value you see in your character sheet.

 

For example, Frost's mastery increases the size of icicles by 2% for each 1 mastery point. So if you have 80% mastery on your character sheet, you'd enter 40, as that's your actual rating. 

 

I think Lhiv did it this way because no one spec converts mastery at the same rating.

[Kver 65]

The up-time of the meta for Frost mages is one of the highest (I think only  Boomkins and maybe Ele shamans have higher?), if you're lucky you can get upwards of 40%. Good idea.

http://www.warcraftlogs.com/reports/mkQAtWqXMbKL7g8V#boss=1623&wipes=1

 

If you're interested in the amount of brain freezes. This is with the 19.4k(ish) breakpoint, and a very high level of snapshotting(If I say so myself). Keep in mind it's progress, don't go tell me that I made a mistake somewhere :D A total of 733 brain freezes, too bad I have nothing to compare it to.

 

It plays like a charm though, and I'd say the DPS I did on Garrosh isn't bad at all, too bad it's not really reliable data to make a conclusion on.

[Ashe 0]

Heres my Compare Bot logs:

 

Immersius H: http://raidbots.com/comparebot/53469c0f74254ebc4000003a

Protectors H: http://raidbots.com/comparebot/5346991874254e9a1500005e

Nourushen H: http://raidbots.com/comparebot/5346995674254e2763000952

Sha of Pride H: http://raidbots.com/comparebot/5346998c74254e796d000450

Nazgrim H: http://raidbots.com/comparebot/53469a7c74254e9a1500006c

Spoils N: http://raidbots.com/comparebot/53469aed74254ebb40000010

Thok N: http://raidbots.com/comparebot/53469b1e74254e9a1500007e

Seigecrafter N: http://raidbots.com/comparebot/53469b4b74254ebb4000001b

Paragons N: http://raidbots.com/comparebot/53469b7b74254ebb40000029

Garrosh N: http://raidbots.com/comparebot/53469bb374254ebc40000014

 

Sorry for missing the missing fights, but logs didnt get uploaded or something happened and they are not

displayed on WoL.

 

Also please not that this is coming from an average mage. Not an elitest raider. I make mistakes (more often than I should...)

So take these logs with a grain of salt. They are also scattered throughout days. I took which logs I had avaible and compared them to

this past weeks kills. All the "After" kills will be the same raid. After is after reaching the 18969 haste point. Everything "Before" is before it.

[Frostedmages 3]

%ETs shouldn't be too difficult to produce. Since you are using your DPET stuff for the damage, you could probably just ignore everything except for haste in game and start casting spells at a dummy over a 5 or 6 minute period a couple times with each haste level.

 

It won't be the best, but it's a good first approximation to how we should expect ETs to change.

[Trailmix 0]

Why does your "FoF+IL avg damage" increase with crit above the crit cap? It should be constant.

 

Also I don't see how you can conclude getting that much haste, especially while acknowledging Mastery is stronger on single-target.

Edited April 10, 2014 by Trailmix

[Frostedmages 3]

Spent 30 minutes (consecutive) swinging on a target dummy with 9606 haste and logged it. I also had 5% spell haste. Spell breakdown by cast:

 

Frostbolt - ~32%

Living Bomb - ~6%

Ice Lance - ~10%

FrostFire Bolt - ~9%

Frozen Orb - ~1%

Welly - ~43%

 

Something was weird with my log in that Welly wasn't showing up properly. I don't know why. He ended up as a separate person from me instead of a damage source, but I'm pretty sure his # of casts is close to correct.

 

I'd imagine as you get higher haste you begin swinging the percents even more in favor of FB, IL, and FFB. I'll need to do some reforging / gemming to hit the ~14k and ~18k haste ratings you have. I'll probably do that tomorrow and Saturday (if I can convince another hunter to chill with me), since swinging at a dummy for 30minutes in a row is pretty boring and I don't want to do it all at once >.>

 

Going from ~9k to ~18k is like +23% haste? Or something close to that, which actually changes the cast time of Frostbolt by a significant amount, like 0.2 - 0.3s per cast? I'll be less lazy and actually figure it out tomorrow morning. Which if I apply to my dummy swings, would mean I'd gain another ~140 Frostbolts, further skewing things towards Frostbolt being an even more dominant source of your damage. Realistically, the increase in frostbolts wouldn't all go into it since you're going to be procing more ILs. However it gets kind of fuzzy past ~14k, since technically every time you are casting IL you are wasting stats that could have been in mastery to empower the frostbolt/FFB it proc'd off it, making the decision between mastery and haste a lot less obvious past that point.

 

I should probably add that I was actually trying to snapshot, so for someone who isn't the ratio is going to be a bit more towards FB/IL/FFB and away from LB.

Edited April 11, 2014 by Frostedmages

[Frostedmages 3]

Why does your "FoF+IL avg damage" increase with crit above the crit cap? It should be constant.

 

Also I don't see how you can conclude getting that much haste, especially while acknowledging Mastery is stronger on single-target.

 

I'm pretty sure it's mentioned in the post somewhere as an edit.

Edited April 11, 2014 by Frostedmages

[Koppi 0]

Should I ever sacrifice a bit of int for my haste breakpoints?

[Muphrid 2]

Spent 30 minutes (consecutive) swinging on a target dummy with 9606 haste and logged it. I also had 5% spell haste. Spell breakdown by cast:

 

Frostbolt - ~32%

Living Bomb - ~6%

Ice Lance - ~10%

FrostFire Bolt - ~9%

Frozen Orb - ~1%

Welly - ~43%

These looks like DPS percentages, not ET percentages. Is that what you intended to provide?

[Kver 65]

These looks like DPS percentages, not ET percentages. Is that what you intended to provide?

43% for the welly. Yes, I suppose. I know that Akraen likes to give the impression but Frosted isn't that dumb :(

[Frostedmages 3]

They are not DPS percentages. LB would be weighted much higher if it represented DPS percentages. Those are the %s for the # of casts of one spell vs. the total number of casts. So Frostbolt would be %Frostbolt = ( #Frostbolt casts / # total casts) * 100

 

I included Welly since Akraen has him in his DPET stuff, so I figured he needed a weight. If you don't include him Frostbolt goes to low 50s and living bomb going up to ~12%. Similar trend though, Living Bomb is a very very small portion of the number of casts you do (which makes sense).

 

If %ET is something different, then drop the forumla and I'll re-do it. I looked around and didn't find anything, so I assumed it was a simple percentage of one spell vs total spells.

Edited April 11, 2014 by Frostedmages

[Muphrid 2]

I see. Well, I would throw the Water Elemental away since it has no effect on the ET%s; perhaps some ET%s can be backed out of your data when some theoretical estimates of cast times are applied. The only big issue I perceive is DoT clipping. Without an estimate of how much time was lost from clipping, it's hard to know what the effective DPET for LB was.

[Frostedmages 3]

I see. Well, I would throw the Water Elemental away since it has no effect on the ET%s; perhaps some ET%s can be backed out of your data when some theoretical estimates of cast times are applied. The only big issue I perceive is DoT clipping. Without an estimate of how much time was lost from clipping, it's hard to know what the effective DPET for LB was.

 

I initially didn't include him, and only did it afterwards because I think Akraen is including him in the final comparison he's doing (could be wrong).

 

It just makes frostbolt worth a ton more, and LB worth very little more. I'll post up what it looks like without him once I get home from work.

 

Not sure how to estimate clipping. You could see how many times I cast it vs. how long it was actually on the target, assume 1 second clipping and then make an estimate using that. But since I tried to snapshot that fucks it up a bit, meaning that I cast more LBs than you normally would if you assume no snapshotting. So the weight given to LB here is actually higher than it should be. You could just use what I have and aknowledge it's an over approximation of the average DPET contribution of living bomb. Really, any sort of weighting will really improve the analysis - the DPET weight of LB is pretty clearly not anywere near that of the other spells. Anything to better represent this reality will be an improvement.

 

I think there are a couple other issues though that DPET doesn't really get at. Haste is being wasted by ~20% of your spell casts (I think it was around 30% without Welly) once you go past ~14k. Meaning, every time I cast IL/FFB, I'm wasting haste due to no longer reducing the GCD. It might begin to be wasted during meta procs for Frostbolt too? Would have to do the math. The haste could have been Mastery on all previous Welly/FFB/Frostbolts, so it's effectivly wasted DPS. Is this taken into account?

Edited April 11, 2014 by Frostedmages

[Muphrid 2]

Sorry, when I said clipping I really meant snapshotting; that's my fault. But what you're suggesting is probably sound: looking at the actual amount of ticks compared to the number of casts should give a reasonable estimate of how much time was lost. On one hand, I'm concerned that stuff like meta going off will skew those numbers, but on the other hand, everybody has a meta, so maybe the numbers should be skewed, so to speak.

I think as far as accounting for haste and GCD capping, DPET should account for this already. If the time to execute an ability is no longer decreasing from haste, and no other stats are being added, then the DPET stays constant even as haste increases.

[Frostedmages 3]

I think as far as accounting for haste and GCD capping, DPET should account for this already. If the time to execute an ability is no longer decreasing from haste, and no other stats are being added, then the DPET stays constant even as haste increases.

 

Not particularly what I was wondering. Say I have ~14k haste. No haste is wasted when I cast IL/FFB, because I have just hit the GCD cap. I calculate my DPETs, and weight them accordingly.

 

Now I move to ~18k haste. ~4k haste rating is now be wasted every time I cast IL/FFB. The DPET due to haste for them has remained constant; however I am losing potential DPET on FFB/WE/FB from the ~4k mastery I could have had. The question is - the is the haste DPET gains for the LB tick and the increased number of Frostbolt/welly hits worth the potetial DPET loss of frostbolt/welly/FFB mastery empowerment?

 

I might be overthinking this and losing myself in my own thoughts. >.>

Edited April 11, 2014 by Frostedmages

[Kver 65]

Not particularly what I was wondering. Say I have ~14k haste. No haste is wasted when I cast IL/FFB, because I have just hit the GCD cap. I calculate my DPETs, and weight them accordingly.

 

Now I move to ~18k haste. ~4k haste rating is now be wasted every time I cast IL/FFB. The DPET due to haste for them has remained constant; however I am losing potential DPET on FFB/WE/FB from the ~4k mastery I could have had. The question is - the is the haste DPET gains for the LB tick and the increased number of Frostbolt/welly hits worth the potetial DPET loss of frostbolt/welly/FFB mastery empowerment?

 

I might be overthinking this and losing myself in my own thoughts. >.>

But that is kind of what we've been trying to work out. I mean, if we didn't cap crit/haste on some spells then it'd be far easier to find a solution. I believe Akraen's research actually proves that haste is stronger for the welly(though what does proof mean at all) 

 

DPET's simply only work up to a certain level, after which they simply leave too many questions. Those are the questions we have now, and if you know a great way of solving that then I'd love to hear it.

 

With regards to the meta, that's a little different. I think you can reasonably well assume a certain degree of snapshotting; we teach people that you always snapshot the meta the moment you get it, and the second before it expires. That means that you can have a reasonable idea of how big the addition of the meta gem is. Of course, sometimes you get a back to back meta gem, which means that you "clip" the uptime because a snapshotted LB lasts 10 seconds longer than the meta gem proc, but it gives a rough idea of the time gained/spent by the meta gem.

 

Of course, all of our research will always be based on rough ideas.

[Akraen 230]

Water Ele DPET is stand-alone. Its ET has no bearing on the caster's ET, and it only casts one spell, so Waterbolt is 100% ET. Sure makes the little fella easy to calculate

 

I'm mulling all this over but I don't think those ET%s are good enough to use. Honestly I'm unsure any are, and an analysis would need to be broken down fight-by-fight.

[Frostedmages 3]

Water Ele DPET is stand-alone. Its ET has no bearing on the caster's ET, and it only casts one spell, so Waterbolt is 100% ET. Sure makes the little fella easy to calculate

 

I'm mulling all this over but I don't think those ET%s are good enough to use. Honestly I'm unsure any are, and an analysis would need to be broken down fight-by-fight.

 

You really need to use something to weight the DPETs accordingly, preferably something derived directly from gameplay.

 

Is it ideal to swing at a dummy? No, it'll only give us patchwerk style ETs; but at least they are real. And honestly, on single target fights, with IF and the number of procs we get the difference in ETs shouldn't be so large that you ruin the trend established.

 

You don't need to break down ETs fight by fight initially. You can establish a baseline of how things work on patchwerk and then use whatever knowledge we have for different fights to change things from there.

Edited April 14, 2014 by Frostedmages