Monday, March 23, 2015

Bloody Stats: Haste vs Multistrike

Death Strike is the Blood Death Knight's bread and butter, so maximizing the number of Death Strikes during a fight greatly increases survivability. In order to maintain a sense of perspective, I'd like to preface this article by saying that skillful play will contribute far more towards this goal than stat optimization. If a DK frequently allows runic power, rune pairs or Blood Tap charges to cap, then the most effective way to increase survivability is to play better.

Also, this only applies to this gear level (673) with this talent setup (not Breath of Sindragosa).

First a few basic facts:
  • DK's level 60 talents (Blood Tap, Runic Corruption, Runic Empowerment) convert runic power into death runes. They are close in performance, but Blood Tap is best.
  • Multistrike auto attacks have a chance to generate runic power via Blood Rites.
  • Haste directly increases rune regeneration.
  • Haste increases attack speed which gives more chances to multistrike, increasing runic power, indirectly increasing rune regeneration.
  • Strength gives DKs parry, parry causes parry haste, which increases attack speed, giving more chances to multistrike, so more runic power, more rune regeneration.

It's largely accepted that multistrike is a great DPS stat for Blood, but some are also claiming that it's also a good survival stat, because multistrike generates runic power, spending runic power regenerates runes, and more runes means more Death Strikes. Additionally, since multistrike is Blood's attuned stat, it gets more powerful the more one has. More runic power spent even increases the uptime of Shadow of Death. It just makes sense to stack multistrike. However, I am wary of things that "just" make sense.

I also see these statements made without any numbers backing them up, and when asked for, variations of "It works for me/It feels right" are the common reply, or sometimes even "Simulations are irrelevant. I don't tank in a sim." (Actually, yes we do.) These are major red flags for me.

I'm a numbers guy, especially for tanking, so I have an appreciation for the work of theorycrafters like Theck and Meloree at Sacred Duty. Their contributions to SimulationCraft have given tanks an easy and accessible method of determining the effectiveness of our stats and abilities. What is shown in these simulations (if used correctly) is as close to objectivity as one can get, and to hear tanks hand-wave this tool and insist on the subjectivity of tanking is disheartening, but this is a discussion for another time. For now I will assume SimCraft to be useful.

My sim results (Result 1 xml here):

Result 1
Fight length 450 seconds
4372 strength buffed, 4164 unbuffed
606 haste from gear (23.28% buffed, 17.41% unbuffed)
672 multistrike from gear (26.44% buffed, 21.19% unbuffed)

112.2 TMI (Theck-Morlee Index, which measures damage smoothing. Lower is better.)
19,880 DPS
30,185.8 HPS
88.2 Death Strikes
91.7 Blood Rites
135.6 Shadow of Death refresh
46.6 parry haste
32.7 Unholy Strength
93.4 seconds waiting (20.8%)

This will act as a baseline to compare two other results too.

Result 2
1,000 multistrike added to gear
(42.35% buffed, 37.1% unbuffed)

106.5 TMI
22,437 DPS
31,346 HPS
93 Death Strikes
147.1 Blood Rites
163.2 Shadow of Death refresh
46.4 parry haste
32.6 Unholy Strength
65.2 seconds waiting (14.5%)

Result 3
1,000 haste added to gear
(36.11% buffed, 29.63% unbuffed)

105.3 TMI
21,217.9 DPS
95.3 Death Strikes
102 Blood Rites
145.1 Shadow of Death refresh
49.4 parry haste
35.9 Unholy Strength
74.2 seconds waiting (16.5%)

These are, I believe, all the relevant factors to rune/runic power generation.

Compared to Result 2, Result 3 is: -1.2 TMI; -1,224.1 DPS (-5.5%); +1,120.7 HPS (+3.5%); +2.3 Death Strikes; -45.2 Blood Rites; -18.1 Shadow of Death; +3 parry haste.

This comparison shows that haste is more of a survival stat than multistrike, while multistrike gives more DPS. While the difference between the TMI scores is small, the DPS and HPS difference is enough to show up on the meters.

In this 450 second time frame, Result 2 (1,000 multistrike) gives 45.2 more Blood Rites than Result 3 (1,000 haste). Each stack of Blood Rites is 1/10th of a Death Strike, so that's 4.5 more Death Strikes more from Blood Rites. Also interesting is that Result 3 only added 10.3 Blood Rites more than the baseline to Result 2's 55.4. This shows that multistrike is working as intended, generating over 5 times as many Blood Rites than haste. However, haste is still the better Death Strike generator overall, and since it isn't generating many Blood Rites, it's through passive rune regeneration.

Result 2 has about 18 more refreshes of Shadow of Death (i.e. 18 more casts of Death Coil) over 450 seconds than Result 3. That's an additional 2.4 every minute. Now, I'm not sure how to directly quantify this increase, so instead I'll compare the total casts of Death Coil between the three results: Result 1 has 136.5, Result 2 has 164.2, Result 3 has 146.1. Result 2 has 121% as many Death Coils than Result 1, while Result 3 has 107%. If this is a fair way to compare the two (and I am uncertain on this point), I think this shows that multistrike is three times as effective as haste in increasing Shadow of Death uptime. I think this is what is putting Result 2 at a similar TMI to Result 3, and is multistrike's actual survival advantage.

Series 1 and 2 have about the same parry haste, while Series 3 is increased by 6%. This correlates with a 10% increased uptime of Unholy Strength. I think this shows that haste increases the uptime of Unholy Strength, which itself increases parry haste, further increasing uptime. That Result 2 is similar to Result 1 shows that multistrike does not increase uptime at all.

Lastly, both Result 2 and 3 have significantly less wait time than Result 1, with multistrike being a little bit better than haste.

In conclusion, both multistrike and haste are good for both damage and survival, but with multistrike being the slightly faster, higher damage stat and haste being the slightly steadier, more consistent survival stat.

No comments:

Post a Comment