Originally Posted by xbit
An argument that I've heard against this is that KM procs make crit less valuable, but the math of that doesn't work out to much. At 5 PPM with a 2.6 speed weapon, KM has a 5 / (60 / 2.6) = 21.67% chance of proccing. Assume that our crit chance is 10%. Crit is wasted whenever we have a KM proc and what would have been a crit anyway occurring at the same time. Since these are independent events, the probability of them occurring simultaneously is simply 21.67% * 10% = 2.167%. So a little over 2% of the time our crit rating is wasted. This devalues crit by 2.167%, but based on the assumption of 40% of our damage being affected by mastery, crit is 20% more valuable. A 2.167% decrease is not enough to give mastery the lead.
Is there a flaw in my reasoning, or should DW frost DKs be stacking crit over mastery?
|
Yes, there are several flaws in your reasoning. First, your PPM is off. Special ability hits and haste increase PPM, and hit rating below white cap decreases PPM. PPM for 1h will probably be from about 6 to 7 depending on gear.
Second, the math just doesn't work. To make things easier, calculations should be done without factoring in the DKs personal crit percentage, as this isn't really relevant if you are just looking for percents. This could be done the way you have it set up, but it gets a lot more complicated and you would have to look at logs and find out how much obliterate and frost strike are actually hitting for, their damage percentage, etc.
This is how it makes more sense to me
Profile used:
Simulationcraft Results (Simulationcraft 372 BIS with my modified action list; this profile is far from BIS for 1h frost anymore, but I'll get to that later)
Note: Including the damage was not necessary for calculations, it is just there for perspective
Total damage = ~9.3 million
9.1 sec interval on killing machine (rounded to 10) = ~40 killing machine procs over the 400 second fight
Obliterate = ~30% of damage (57k crit, 27.5k hit) with 74 uses over 400 seconds
Frost Strike = ~25% of damage (33.7k crit, 16.4k hit) with 108 uses over 400 seconds
Estimate – 20 KM procs used on both Obliterate and Frost Strike; 20/74 = 27% & 20/108= 18.6%
Obliterate – (27%*30%*9.3mil) = 8.1% or 753k damage
Frost Strike – (18.6%*25%*9.3mil) = 4.7% or 473k damage
8.1% + 4.7% = ~13% of the time crit rating is not effecting our damage in any way
This means we are benefiting from crit rating 87% * 1.03% = ~90% (1.03% is from 3% crit damage meta). This only applies to this specific encounter with this gear. This will vary from playstyle, gear, and spec, but is should be roughly around that mark for most people.
Now, that might make you think that crit is automatically better, but I wouldn't be so sure. First, the percentage of frost damage dealt each fight will vary greatly from fight to fight and from player to player. The fact is, in every single raid encounter except for Chimaeron and Atramedes, you will be dealing some form of AoE frost damage (or you should be). However, only Chimaeron would really count as a single target because when Atramedes takes off, you should still be using howling blast and have dots ticking. This means that on every fight except for Chimaeron (if you are playing right), your actual value of mastery should be from ~2%-40% (extreme AoE fights) higher.
For example, on V&T most DKs probably don't keep dots rolling on both dragons (one on ground, one in air), which not only increases DPS, but also increases the value of mastery. In my most recent V&T kill, frost was over 45% of my damage...and this is as 2H with under 1k mastery.
Most recent V&T kill:
World of Logs - Real Time Raid Analysis
Most recent Nef kill:
World of Logs - Real Time Raid Analysis over 45% of damage is frost as well
For both of those kills my stat priority was haste > crit = mastery with a 358 ilvl. I generally switch Impatience of Youth (mastery) for Fury of Angerforge (crit) depending on the level of AoE (I think I was using crit for both those fights). I was running with ~1450 haste, ~900 crit, ~900 mastery.
Obviously, everything I am saying pretty much goes against the general consensus on DK. I sim 2h frost and 2h unholy at almost identical levels at 359 and both specs benefit by far the most from haste. UP scales better with haste, so with enough haste (well above 359) it will become the clear choice; so for the vast majority of raiders, stacking haste in FP for 2H is the way to go.
Scale factors on my own 358 profile:
Simulationcraft Results 1H
Hit to spell hit cap: 1.04 (every 1H DK should be getting 10% hit)
Crit: 0.82
Haste: 1.00
Mastery: 0.725
Simulationcraft Results 2H
Crit: 0.73
Haste: 0.78
Mastery: 0.696
(Just because the factors are lower for 2H than for 1H means very little, as increasing one stat effects others. This just means you get a higher value for hit and expertise as 2H FP as you are more GCD constrained than 1h frost.)
I simmed it with my mastery trinket on. My single target would increase with the crit trinket. However, the more I look into the numbers the more convinced I am that AoE frost damage on most raid fights is a lot higher than most would expect. So, even though mastery scales slightly slower than crit (very close for 2H, surprisingly frost does less % damage for 1h and is a bit weaker), if you are properly rolling dots (like you can on 11/12 raid fights) then mastery is likely to overtake it, especially with how close the stats are for 2H. It's a bit more of a tossup with 1H - I really don't believe you will be gimping yourself by stacking mastery over crit (as long as haste is first priority) though as they should be close. And, I hope it goes without saying that on a more intensive AoE fight like Halfus, Magmaw, or Maloriak, mastery wins by a mile. One final note as well; in the simulator switching stances is still consuming a rune, so this will increase the damage further for both of my specs, as they both involve swapping during bloodlust. Human players can again increase DPS even further in actual practice doing this by being intelligent about dumping RP right before and after bloodlust (I have it set up to do this to some extent, but a player could be far more efficient).