Scaling by spec

[Cheeky]

There is a lot of talk on the offical Hunter forums about how once you get into T5/6 content that MM "scales" better than BM and becomes a better raiding spec. I've always been dubious of those claims, but MM/BM is a religious war I avoid on the official forums.

But, my spreadsheet now tells you how much each stat helps in terms of DPS for your spec/gear, so I figured why not kill an hour while waiting on a massive compile to put the myth to the test? I'm not going to post this on the official forums because I don't want to deal with the insane amounts of crap people would add. TKA is a bit more "enlightened" in handling arguments.

I'll present all the data, and what I also saw playing around. A summary at the end.

Caveats:
1 - This is based of the math and models in my spreadsheet being correct. If you already think they are wrong, then please stop and read another thread. There will be nothing to interest you here.

2 - Stand-still DPS. No movement, no breaks, no mana issues.

3 - Level 73 target with 35% armor reduction.

4 - Race is Troll (I had to pick one, and I got halfway through before realizing I picked one with an applicable racial bonus. Oh well.)

5 - Latency is 0.2s (mine)

6 - The following raid buffs were used: Kings, Might, LotP, AB, GotW (Improved 5/5), Agility Scroll, Agility food, Elixir of Major Agility.

7 - BM used a 1:1, replacement rotation. MM used a 1:1.5 without IAotH.

8 - Cookie cutter talent specs 41/20/0 and 0/4x/y were used.

Gear sets (I picked gems that made the most sense. Not always the same between specs):

Slot                Tier 4                Tier 5                   Tier 6
-----------------------------------------------------------------------------
Melee               Legacy                 Twinblade         Halbard of Desol
Head                Demon Stalker          Rift Stalker      Gronnstalker
Neck                Worgen's Claw          Choker of Fury    Endless Nightmares
Shoulders           Demon Stalker          Rift Stalker      Gronnstalker
Back                Vengeance Wrap         Thalassian        Shadowmoon Destro
Chest               Demon Stalker          Rift Stalker      Gronnstalker
Wrist               Stalker's War Bands    Celestial Archer  Percise Flight
Hands               Demon Stalker          Rift Stalker      Gronnstalker
Waist               Girdle of the Prowler  Black Eagle       Boneweave
Legs                Demon Stalker          Rift Stalker      Gronnstalker
Feet                Fiend Slayer Boots     Cobra-Lash        Quickstrider
Ring                Garona's Signet        Lethality         Eternity
Ring                Ring of Marks          Ring of Marks     Lethality
Trinket             Hourglass              Tsunami Talis     Madness Betrayer
Trinket             Bloodlust              Bloodlust         Ashtongue
BM Weapon           Wolfslayer             Wolfslayer        Wolfslayer
MM Weapon           Sunfury                Serpentspine      Betrayer

First set of information, basic stats:

                   Tier 4         Tier 5        Tier 6
                 BM      MM      BM     MM     BM     MM 
---------------------------------------------------------------
Health          8824    8404    9587   9130   9910   9438
Mana            8614    8945    8499   8822   8812   9155
Armor           6827    6856    7455   7408   8063   8096
Tooltip RAP     2212    2780    2440   2991   2551   3148
Tooltip Crit   34.35%  34.52%  39.26% 39.51% 41.22% 41.22%
Hunter DPS      1004    1030    1088   1171   1175   1264
Pet DPS          317     191     327    197    333    200
Total           1321    1221    1415   1367   1508   1464

And what you'd see based on one more of each stat:

              Tier 4        Tier 5           Tier 6
Stat        BM      MM    BM      MM       BM    MM
------------------------------------------------------
Agility   +0.560  +0.542  +0.570 +0.573  +0.587 +0.596
AP        +0.260  +0.235  +0.265 +0.251  +0.277 +0.259
Crit      +0.450  +0.448  +0.458 +0.469  +0.465 +0.491
Hit       +0.438  +0.000  +0.452 +0.000  +0.000 +0.000
Int       +0.000  +0.123  +0.000 +0.132  +0.000 +0.136

First Conclusions:
Well, they are right that MM starts to scale better at higher gear levels. But this is 100% due to the fact that the Wolfslayer Sniper Rifle remains the best BM weapon in a 1:1 rotation for the gear listed. Not being able to upgrade your ranged weapon is a bit of a problem. You'll also notice that pet scaling is really pretty poor. You don't even see a +10% gain from T4 to T6. Tier 5 gear seems to be the biggest jump for MM spec, and that's because you can really take advantage of the powerful, slower weapons.

Also, Agility starts off being the best stat, and only gets better as gearing improves. While this isn't a surprise to anyone that's done some number crunching I was impressed on how quickly it leaves everything else behind. (I specifically avoided haste so far.)

BM and it's saving grace statistic
I then said, "Well fuck, if BM is getting punished by how slow the weapons are, what if we speed them up?" and made the following gear substitutions:
Black Bow of the Betrayer
Dragonspine Trophy
Fists of Mukoa
Bindings of Lightning Reflexes

Here are the results:

Health          9968
Mana            8548
Armor           7983
Tooltip RAP     2478
Tooltip Crit   37.46%
Hunter DPS      1236
Pet DPS          330
Total           1566

Scaling for the BM/Haste build:
Agility +0.619
AP      +0.283
Crit    +0.507
Hit     +0.000
Int     +0.000
Haste   +0.517

Whoa! BM just picked up its same advantage it enjoyed in T4 gear, and the Hunter himself isn't doing too much less damage (2.2%) than the MM Hunter in pet-unfriendly fights. And if you look at the scaling numbers, that gets better and better as you gear up in Sunwell and beyond.

So I think (barring a mechanics change) I'm going to stay BM while raiding and salivate over the slower weapons and haste gear. (And I need to get a DST now, to bank for 6 months from now when I'm well into BT.) With a 1:1.5 rotation, MM specs see little benefit from haste, the weapons are just perfect for the rotation right now.

Please feel free to critque my method or results. I didn't spend the time to check each and every piece of gear available at each level, there may be better combinations that I ignored. I think this is a good baseline to use. I may post more conclusions as I think about this more.

[Seipher]

Pretty interesting stuff. I am currently doing BM with the Merc. Gladiator bow. It has been my observations while actually testing out MM, Surv and now BM. that really isn't accurate to model latency as a static average value.

For example if you have a fixed 0.2ms average latency you will never clip shots in the spreadsheet, however realistically that latency has some jitter to it (sometimes a lot of jitter) and so for some shots you will hit high latency and clip a shot. Let me know if I am wrong here, but I was thinking it would be sensible to introduce a distribution to the latency so that you could modeled the proportion of shots you clip due to jitter. I think it is that will probably have an effect on average MM dps as well as on weapon selection for BM.

[Cheeky]

Originally Posted by Seipher

Pretty interesting stuff. I am currently doing BM with the Merc. Gladiator bow. It has been my observations while actually testing out MM, Surv and now BM. that really isn't accurate to model latency as a static average value.

For example if you have a fixed 0.2ms average latency you will never clip shots in the spreadsheet, however realistically that latency has some jitter to it (sometimes a lot of jitter) and so for some shots you will hit high latency and clip a shot. Let me know if I am wrong here, but I was thinking it would be sensible to introduce a distribution to the latency so that you could modeled the proportion of shots you clip due to jitter. I think it is that will probably have an effect on average MM dps as well as on weapon selection for BM.

Good god. While you might be right in what actually happens there is no real way to model that. My experience has been that the jitter to latency is small compared to the latency value itself, but that may be because I'm on a decidicated FTTH connection with excellent bandwidth. Others may see much worse stability.

In terms of using the spreadsheet I'd just go with an average value.

[jinserai]

Yeah, latency jitter is a HUGE problem with my cable connection (although it's more reliable using DSL, it's reliably worse). Trying to model something like this would simply be impossible, as it not only varies from connection to connection, but also day by day. Some days are really spiky, while others are semi-constant.

Incidentally, rabid latency times are why I went BM in the first place. There aren't any 2.4 speed weapons after the consortium Blaster.

[Arewhen]

As the Systems Architect for the equipment that runs FiOS - Cheeky, you are welcome.

[Seipher]

Actually I don't think it would be that hard. I can probably write a mod something to collect actual in game latency statistics over time. Generally, in the long run it should be reasonable to model the jitter by measuring the average latency and the variance and using a Chi-squared distribution (perhaps someone has a better suggestion). A mod could be written like say performance_fu to display latency variance in addition to the latency average which players could plug into the spreadsheet.

It would then be possible to calculate what proportion of shots will be clipped and even determine what proportion would be clipped by 0.1-0.2s, 0.2-0.3s, 0.3-0.4s, etc. A simple histogram approach by binning the number of clipped shots for small intervals until reaching an interval that is less than 1 in 100 or 1000 would probably be sufficient.

Let me think about it a little and get back to you. If I can get a decent model of shot clipping this way would you be interested in testing it out?

As for my own connection, I have a fairly reliable cable connection (most of the time) that has excellent throughput and I get low latency with most other gaming servers. Doing ping tests and traceroutes doesn't show any problem before I start hitting the Blizzard servers. I should probably do some more thorough tests but I am sure everyone is affected by jitter during peak times. That said I am in Canada and I am sure distance is a factor.

I would think that in general, since internet connections in general are inherently bursty, large, though not noticeable to the player, latency spikes are happening all the time. You may not see them but they probably clip shots for any tight rotations.

[Cheeky]

Originally Posted by Arewhen

As the Systems Architect for the equipment that runs FiOS - Cheeky, you are welcome.

Cool. If I provide some TheoryCrafting for you, can I get a discount in my rates, or free HD TV?

You on the access side or transport? I build metro transport ROADM equipment myself.

[Luciellena]

You said you used a 0/4x/y spec for MM. What about trying a y/4x/0 spec to boost your pets DPS some?

I would think it only fair to compare a max DPS MM build against the 41/20 BM build. Something like the following perhaps:

http://www.wowhead.com/?talent=cVbR0xhZVqbRVzhest

[Cheeky]

Originally Posted by Luciellena

You said you used a 0/4x/y spec for MM. What about trying a y/4x/0 spec to boost your pets DPS some?

I would think it only fair to compare a max DPS MM build against the 41/20 BM build. Something like the following perhaps:

http://www.wowhead.com/?talent=cVbR0xhZVqbRVzhest

I was looking to do a quick (took over an hour) compare. I don't really want to go through all the permutations of different talents. The spreadsheet is public knowledge, and anyone can try it out for their favorite gear/spec.

[Arewhen]

Access - I was the AFC CTO before it was acquired by the 10th Circle of Hell aka Tellabs. Now, I am just a Strategy VP for Tellabs.