Thank you for the sim results and helpful answers Asphyxialol. That about clears it all up for me.

Re: Syrellia, I personally take subversion not for the threat reduction but for +crit bonus. If not for the 3 points in subversion, however, where would you spend them in a non-improved icy talons build?

If you pay attention to the DPS in the simulator, the last 35 hours of the simulation (or 35% of the simulation) your OB / FS / HB are affected (as shown by the slightly increasing DPS that casually goes up the last 35% of the calculation). Or, alternatively, you could simply run a standard test *without* 2 points spent anywhere (which we can use one of the above examples for) and do the math by hand yourself. You would factor in that 35% of the overall damage is affected by 1.12x the damage to receive the additional damage output from MC and then it would come out to the same calculations as above.

From the above examples the 3/51/17 set will not have MC and the addition of BCB has no effect on the difference in damage from MC.

So, MC affects IT, OB, HB, and FS from 35% of target life to death and multiply their existing damage by 12% (1.12). From my first test these were the values -
Obliterate - 731235102
Frost Strike - 555532060
Howling Blast - 137896021
Icy Touch - 61251948

You take 35% of the overall damage and then multiply that by 12% and then subtract the raw 35% value to find the increase in overall damage done... as an example for OB:

731235102 * 0.35 = 255932285.7 * 1.12 = 286644159.984 - 255932285.7 = 30711874.284 increase in total damage from MC, for a total of 761946976.284. This alone is is a 1.0123% increase in DPS, before factoring in KM/Rime HBs, IT, or FS. Doing *very* raw napkin math (more or less based off of percents because I'm lazy), FS would be 0.767%, HB would be 0.19% and IT would be 0.0864% with MC, totaling 2.0557% in DPS (where, in my example, BCB was worth 1.8%).

Obviously these value change based on rotation (as an example I run GoDisease, thus Armor Penetration is worth more to me than a typical IIT build) and whether it favors haste or arm pen, or whether you have 4pc t8 or not.

Regarding Subversion vs BCB and its affect on sigil vs 4pc t8, this is true, but consider the extremely small difference it makes in a parse. If I remember right the math showed that the 4pc bonus on a blood build did roughly 1-2% difference in DPS when the strike it affected pulls about the same as OB+BS in most IIT builds, making it a truly insignificant difference in actual output. Regarding sigil - the math has already been done showing the difference between Virulence and Awareness being so insubstantial that most DKs are simply continuing to use Awareness and getting the sigil last.

I ran two more parses now, one as 0/53/18 and one as 3/53/15, both without the 4pc bonus, and the DPS I saw was 6798 from 0/53/18 and 6835 from 3/53/15. As I said earlier the difference was 15 dps lower using Subversion, now it is 37 dps more, mostly due to the stat upgrades (not that I have a lack of haste, I sit with around 230 haste and I believe 226 armor pen not factoring in Mjolnir).

As I said though, when it comes to ~30 dps, the damage difference is so minuscule that it is hardly worth considering as RNG alone could sway the fight by a means of a couple hundred DPS to begin with. What matters is the passive bonus you get. Considering I don't have UA and thus I am entirely reliant on procs the whole 'use all your cooldowns during 35% to get the biggest benefit' doesn't really apply to me. Also, I don't believe there is a single fight we use Bloodlust late with the exception of Algalon, or Steelbreaker (which is only late comparative to the fight, not to his personal health), which again just puts the math in my favor of BCB over MC.

Also, in regards to the post you made while running the sims for my post (the threat post) - decent tanks should be able to consistently out-threat a dk pushing about 5-6.5k threat? I will refer you to this post: Premonition - Sen'jin (US) - View Single Post - 3.2 Tank Tests

That threat compares (from the beta, mind you) the TPS generation of the 4 tank classes in 3.2 on Patchwerk. The Patchwerk dummies on the PTR are arguably the best means to gauge TPS and DPS considering there are no interrupts, no raid wide damage, no movement, nothing. It is simply stand there and blow everything you have to kill it. In these examples the two classes with the least TPS would have issues holding aggro over a dk without subversion (druid and dk). The warrior (Xav) is considered to be one of the top TPS warriors in the world, if you've ever seen one of his videos you would know why, and he could even have issues variant on the RNG of his abilities. The only class which could consistently keep threat over you despite your spikes, and allow you to start in the beginning of a match and have 100% presence during the encounter is the paladin tank. The addition of subversion to your build allows you not only to start early into an encounter, allowing longer dps time and a higher percentage of presence on the fight, but it allows you to have the cushion necessary to spike your damage during a period of high procs or lust, etc.

Now, personally, even with subversion, I have been known to pull in the beginning of the fight due to a string of crits, or heavy procs at the start of a match, or poor RNG from my tanks, etc. and I greatly value the cushion of room later in the fight when I have my maximized dots ticking, and MC is active, and I have been riding the tanks threat for... a pretty good portion of the fight on most fights (obviously excluding stand still and nuke oriented fights like XT or Vezax).

I run a 2/51/18 GoD build and decided to compare the % dmg BCB yields compared to Subversion. I compared my logs of multiple iron council hardmode wipes because its pretty much a single target dps fight, there are some times when I am in a rune of power but I don't think it would skew a comparison between subversion and BCB.:

World of Logs - Real Time Raid Analysis

World of Logs - Real Time Raid Analysis

I believe I was blood spec in a few of these. In the second log attempts only 1 and 9 are frost. There were a total of 9 attempts that I took my averages from, attempts 1-7 from log 1 and attempts 1 and 9 from log 2. I disregard the kills because that was not done in hardmode.

I would take my total BCB dmg from each attempt, divide by 3 to get the %dps increase gained from 1 talent and I think it averaged out to be around 0.96%.

For subversion I'm not totally sure if this is right, for 3% crit increase on obliterate but since obliterate bonus is actually 145% I multiplied 3% crit increase by 1.45 and got 4.35% as my factor. I then totaled my obliterate damage and multiplied by .0435 in order to get the % dps gain from 1 point in subversion. By this calculation I got 1.63% dps increase on average.

In conclusion, 1 point in subversion yielded 1.63% dps and bcb .93% dps. So I for me I should probably go 3/51/17. I am a little worried about how I calculated the % dps increase for subversion it seemed really high, and my BCB seemed really low. If anyone has any comments or concerns on that please let me know.

When today's log gets posted I'll have a little more to work with, I did about 3 attempts as blood and 3 as frost. I think these new ones should be more accurate because I'm perfecting my rotation more with each attempt.

My parses show that, as DW, BCB does anywhere between 1.6% and 2.1% of my damage on any given single target fight with 2/3 (comparing XT, Assembly, and Jaraxxus primarily).

As an example here is my Jaraxxus log - WoW Meter Online - Combatlog Replay

The log itself is a very solid log, I believe I was on Jaraxxus pretty much 100% of the fight and the damage is not inflated by hitting other available targets (1,714,712 to Jaraxxus, 41,084 to Infernal and 13,466 to Mistress of Pain - ie approximately 3% of my personal DPS was to something other than Jaraxxus). Keep in mind the percentages on the parse itself are incorrect as it factors in ghoul and AotD, so I will break down the BCB percentage for you...

1,724,793 total damage and 36,314 BCB damage = 2.1% BCB damage, or 1.05% per point, and this is *very* slightly off from a full single target, where it would be about 1.08% on a pure single target, but due to the nature of the fight diseases / HB had hit other targets.

I will take the standard hit / crit from that log to use as my data when doing some rough calculations. I will figure crit without Subversion as 56% and crit with Subversion at 65% and use 100 MH and OH strikes to keep it even.

Avg MH Hit: 3818
Avg MH Crit: 8903
Avg OH Hit: 2382
Avg OH Crit: 6096

Damage w/out Subversion: 1,112,744
Damage w/ Subversion: 1,191,935
Difference: 79,191

This is a gain of 7.12% on OB. In my GoDisease parse OB was worth 38.6% of my overall personal dps (ie, no AotD / Ghoul), making the effective boost of 7.12% worth 2.74832% of my DPS, or about .916% per point vs the 1.08% of BCB, a difference of .164% dps per point in a fight with virtually no interruptions to my rotation.

Yeah, I understand what you're saying. The static increase of raid dps within execute range would decrease the actual amount of time spent in execute range so that the time spent was actually less than 35% of the time.

I'm not sure if the simulator takes that into account, but even if it doesn't it would assume the best circumstances for personal dps sub-35% if it used my method, so if BCB outshines it then, it would perform better in a raid setting in actuality.

I see, I recalculated 1 point of subversion from my first on log over all attempts for simplicity.

Given:
375.% damage from obliterate
465 landed obliterates (MH+ OH)
avg MH hit: 3,448
avg MH crit: 8,967
avg OH hit: 2,305
avg OH crit: 6,040
total obliterate dmg dealt (2/3 subversion): 5,173,322
crit chance for obliterate (2/3 subversion): 57.31%
crit chance with 3/3: 60.31%

In my calculation I added MH and OH hits and crits together respectively, so avg hit was 5753 and crit 15,007.

avg hit/crit * 2/3 subversion crit factor
5753 * (1 - .5731) = 2,456
15007 * .5731 = 8,601

avg dmg per cast of Oblit(2/3 subversion): 11,048.8

avg hit/crit * 3/3 subversion crit factor
5753 * (1 - .6031) = 2,283
15007 * .6031 = 9,051

avg dmg per cast of Oblit(3/3 subversion): 11,334.1

multiplied each by the number of casts of obliterate:
2/3 subversion: 11,048.8 * 465 = 5,137,692.0
3/3 subversion: 11,334.1 * 465 = 5,270,356.5

difference: 132,664.5

from my log my obliterate dealt 5,173,322 damage

132,664.5/5,173,322 * 100 =2.56% dmg increase for obliterate

2.56% * 37.5% = .96% dps increase for 1 putting a final point into subversion.

OKAY so using your method on my numbers I got something that looks a lot like what you calculated. I see how this works now and its a lot more complicated than I thought. So for me putting the extra point in subversion wouldn't really increase my dps much at all, I would however gain additional threat reduction.

The next thing I'd like to look into is if dropping 1 point in BCB for Unbreakable Armor would be worth it, especially with GoD and disease refreshing. Maybe I'll run the simulator with a 3/52/16 spec. I'm not super familiar with Kahorie's DK simulator with using GoD though.


[edit] I'm still a bit baffled as to why my BCB dmg is very low

[edit2] Looking over what I had done previously to calculate 1 point in subversion by multiplying the obliterate dmg by .0435 and getting 1.63% dps increase, I think doing it this way assumes you have zero % crit chance. So taking my 1.63% dps increase and multiplying by .57 which was my obliterate crit chance I get .93% which seems a lot more reasonable and closer to what I calculated in this post.

Bear in mind that although many classes have execute-style effects, very few of them make a significant difference to dps. Notably, [frost]fire mages and affliction warlocks have powerful execute effects, in the form of molten fury and death's embrace/drain soul, respectively, which provide a vast increase to their low-health dps. Execute isn't a significant increase for warriors, and talents like decimation and dirty deeds are essentially non-factors as they don't appear commonly in raids. Hammer of wrath and kill shot provide moderate increases for ret pallies and hunters, I believe, but the dps increase is hardly on the scale you propose. Obviously this will vary with your raid comp, but I'd be surprised if the average raid's dps increased by even 5% between 80% and 20%, so you're talking about maybe 33.9% of a tank-and-spank fight during which MC is active, or 3.3-3.4 min in MC range given 6.5 min out. Your numbers claim that rDPS increases by 75% when the boss hits 35% health, which is excessive to say the least, even if you assume that heroism is always popped around 20%.

Regardless, Merciless Combat's greatest worth is not in its averaged-out dps value, but the additional value it provides when low health is the important burn phase.

I'm still keeping my opinion about threats. I don't understand why your reference would make it any relevant, since in patchwerk-like fights, those tanks are doing 7 - 10k tps which is massively more than 5k tps we're producing as predicted in the simulator in patchwerk-like fights. In addition, let's do a little bit of analysis:

The rotation always start with:

IT > PS > OB > BS > BS > FS > FS (the first 10s)

Both IT and PS will produce very low threats. After that you have OB to have a chance to give massive threats. Then 2x BS will do low threats as well. So basically you're giving the tank at least 3s to build up aggro before commencing on the first high threat attack, and 10s to build up aggro in total before nuking as hard as you can with the 2nd part of the rotation: OB > OB > OB > FS > FS > FS. Honestly I don't see why the tank can have any problem with holding the aggro, as I myself have never been able to produce massive threats at the start of any fight. Because of the way we build up our diseases before the fight starts, we will never be able to produce a lot of snap aggro at the beginning of the fight. However you are right, the threat reduction will always put you in low threat category which should not be discounted.

Regarding BCB vs. Subversion: It was producing 50 dps difference per 3 points in total when I run the sims without sigil of Awareness but with sigil of Virulence and 4xt8. 50 dps is not significant, but my concern is once tier 9 comes into play and we abandon tier 8, the dps difference could get even wider.

The mistake you made here is the same mistake most people make.

3% crit is NOT 3% dps increase.

The only case in which that proves to be true, is when you have 0% crit to start with.

So it totally depends on your critrate how much of an increase subversion is. But with 60% crit it should indeed be around 0.9% dps.

However BCB is very unreliable, I have seen it spike at 4% of my dps on some fights, and dip as low as 1.2% of my total dps on other fights (yes both hands combined).
I personally think that 3/53/15 is a much better spec.

@Syrellia:
Don't fully agree there.
Although I agree it might not matter on patchwerk like fights where the tank stands and goes up to 8k TPS. For most fights in the current game that situation does not hold up.

I personally value the threat reduction from subversion relatively high.
Several boss fights have some dodgy mechanics where threat can become an issue.

BCB is unreliable, and you can see it visually. Subversion is unreliable in the same way, but you can't see it visually unfortunately. No matter what you do, RNG won't allow you to crit at the same rate with OB in every single fights. RNG behaves the same way for BCB, so sometimes it procs more sometimes it won't.

For AoE fights, if you count in all the AoE you've done, surely BCB won't make up much of the portions in your dps (you can see it by checking parses). However, the % of which Obliterate counts toward your dps will be much lower, hence the effectiveness of Subversion is much lower - but you can't see it visually when checking parses because there simply are nothing you can see from Subversion. If you're using runes to cast Howling Blast to AoE, surely Subversion will be less valuable because you're trading Obliterate for more AoE damages from Howling Blast. BCB is not affected by such.

I do agree that in some fights where melee get jacked up dps, tanks may have trouble holding aggro (I could only count in Hodir so far, where melee may get buffs while tank doesn't). However in a mobile fight, a decent tank would be able to keep 90% of aggro despite the movement, while the melee cannot do 90% of his dps on boss (a tank would be on boss more than a melee dps would be). Hence, the value of threats the melee dps produces will be lower as well. If a decent tank can hold up aggro in a patchwerk-like fight, I don't see any reason why he cannot hold up aggro in a mobile fight as well.

If you're talking about Kahorie's DK simulator, it says in the read me file that
"* Merciless Combat is active 35% of the time simulation"
So the simulator doesn't calculate any type of diminishing returns on that.

@Foxx2405 the 3/53/15 spec you suggested, does it include improved icy talons? or deathchill/UA? If I wanted to pickup improved icy talons and UA what would be the best talent to drop for UA, probably 1 pt in necrosis?