Carnage/Combat stat weight spreadsheet online

[LagunaD]

I have posted a spreadsheet on Google to help Carnage Marauders and Combat Sentinels optimize their stats.

The link is:

http://tinyurl.com/89d27x5

Working with the spreadsheet

To get an editable copy of the spreadsheet (so you can enter your own stats), you have two options:

1) Log in to Google Documents using your Google/Gmail userID (or register, if you don't have one), then choose File->Make a Copy... from the spreadsheet menu bar. The copy will be placed in your Google Documents, and will be owned/editable online by you. This option requires a Google login but does not require any spreadsheet software.

2) Export the spreadsheet to a format of your choosing (e.g. Excel) and download it to your computer by choosing File->Download As from the spreadsheet menu bar. Once the file is downloaded you will be able to edit it on your computer. This option does not require a Google login, but does require compatible spreadsheet software on your computer.

Once you have a private copy, for most users it will be sufficient to enter your character's stats in the green-shaded fields on the first page ("Stats"). The stat weights are displayed on the same page, along with a simple "Upgrade Advisor".

The target's characteristics are guessed on the second page ("Defender"), and can be modified if desired.

Data in grey- or red-shaded fields should be considered "Read-Only", and should not be modified.

The model

The spreadsheet is only valid for Level 50 characters, and assumes a standard build (31+5+5). The usual +5% group buffs are assumed.

The model assumes that Battering Assault, Gore, Ravage and Force Scream are used on cooldown, and adjusts the rates of Assault and Massacre to create an average Rage-neutral "rotation" of abilities.

The expected rate of Rage generation from Blood Frenzy is calculated and factored into the rate of Massacres and Assaults.

The model assumes Force Scream is only used when the auto-crit buff from Blood Frenzy is up.

The model assumes Berserk is used immediately when 30 stacks of Fury are accumulated. In practice, Berserk is treated as a second Rage-neutral rotation which is used for the calculated fraction of the time that the Berserk buff is active. The modified cost and GCD of Massacre is accounted for during this Berserk rotation.

The rate of random Ataru procs is determined using the calculated uptime of the Massacre buff.

Gore is treated as a reduction of the average enemy mitigation, reflecting the fraction of the time the armor penetration buff is up.

The Execute buff is treated as an average bonus to all rage-spending abilities, based on the calculated uptime of the buff.

It is important to note that the ideal "rotation" used is almost certainly not achievable in practice; nevertheless, it should model the ability selections of a skilled player - in a long fight - closely enough to calculate useful stat weights.

The stat weights

The stat weights are the derivatives of average DPS with respect to each stat. For convenience, both the raw derivatives and the values normalized to (Strength = 1) are displayed.

The general procedure for calculating stat weights is described in another thread, here: http://sithwarrior.com/forums/Thread-Pri...at-weights

A build is "optimized" when all stats with at least one point have equal weights. Non-optimal builds can increase their expected DPS (and thus become closer to optimal) by moving points from stats with low weights to stats with higher weights.

Another approach is to use the Excel Solver tool on a downloaded copy of the spreadsheet, constraining the total stat budget to be fixed and maximizing the expected DPS with respect to the four secondary stats.

Discussion

The model shows that, at least up to raid-entry level gearing, Power and Surge Rating are the most effective secondary stats.

The value of Accuracy Rating is somewhat reduced by its irrelevance to Force Scream and weak effect on Ataru procs.

The value of Critical Rating is likewise somewhat reduced by its irrelevance to Force Scream.

The weights of Accuracy and Critical Rating are close enough to Power and Surge that they do not dramatically decrease expected DPS (at least in moderate amounts) and may become competitive with Power/Surge with higher gear levels.

Edit: February 10, 10pm PST
Updated with Force Choke and the 1.1.3 Surge nerf

[Ezmode]

I was going to ask you to take a look at a 100 seconds worth of a Carnage rotation, but I can't PM you the link. Essentially, I used Ravage and Force scream only after Gore, BA 1 GCD before Gore. Easier if you just take a look at it.

http://www.mediafire.com/?nu3z9t7x5vozycx

[LagunaD]

(02-05-2012 01:42 PM)Ezmode Wrote:  I was going to ask you to take a look at a 100 seconds worth of a Carnage rotation, but I can't PM you the link. Essentially, I used Ravage and Force scream only after Gore, BA 1 GCD before Gore. Easier if you just take a look at it.

http://www.mediafire.com/?nu3z9t7x5vozycx

Not sure what your point is.

Re: Force Scream + Gore

If you use Force Scream "only after Gore", you are using Force Scream at most 4 times per minute, compared to the maximum of 6.67 times/minute the spreadsheet assumes.

If Force Scream's crit damage is Q, and the boss's mitigation is X, your Force Scream will do 4*Q per minute.

In the spreadsheet's calculation, Force Scream will do (20/3)*(1-0.7*X)*Q, where 0.7*X is the average mitigation taking into account the Gore buff's uptime.

20/3*(1-0.7X) > 4

when X < 57.1% mitigation, which I think will usually be the case, particularly in a raid where other classes will be stacking armor debuffs.

Putting it another way, Gore removes mitigation for 3 GCDs out of every 10 (plus the GCD where Gore itself is used). That means in my model Force Scream is used after Gore (20/3)*(3/10) = 2 times per minute, and used without the Gore buff 4.67 times per minute. For your suggested usage pattern to be superior, 2 crit Force Screams without mitigation would have to be better than 4.67 crit Force Screams with mitigation. Again, I can't see how that would ever be the case.

Re: Battering Assault + Gore

Using Battering Assault "1 GCD before Gore" means you are using BA a maximum of 4 times per minute (or using Gore less than 4 times per minute, or both), compared to the 5 times per minute in the spreadsheet's calculation. This will reduce your Rage generation by 6 per minute (out of a theoretical total of about 51 per minute), for about 12% less Rage generation, requiring more Assaults. Or fewer uses of Gore.

Re: Ability selection in general

With that said, the spreadsheet can be modified to work with any specified pattern of skill selection, but this becomes very complicated. The cycle of cooldowns repeats once every 3 minutes (=270 GCDs). But the Rage return from Blood Frenzy can't be predicted at the level of individual GCDs, and Berserk will occur several times during that interval to complicate things further. In any real fight, there will also be voluntary and involuntary movement, CC, breaks in the fighting, etc. which even a simulator can't anticipate.

Generally, we want to keep cooldowns on cooldown, and generally we need to be Rage-neutral. That is why the spreadsheet looks at average frequecies and imposes average Rage-neutrality.

The purpose of the spreadsheet is not compare different ability priorities or rotations, however - for that, you need a full combat simulation that includes the random nature of Blood Frenzy regen. It can probably model things like using or not using Ravage at all, but the primary purpose is to calculate stat weights, and the exact frequency and pattern of ability use is a second order effect on those. As long as the proportion of melee damage to force damage is roughly the same as the spreadsheet calculates, the stat weights will not be much different.

[Ezmode]

Thanks for the explanation, I'll make changes based on that. A couple questions, how hard would it be to readjust your spreadsheet for the sake of calculating for Annihilation, Rage, or any other Warrior tree? And, ill you be taking Vicious Throw into account at any point?

[Bane]

Awesome work!

[LagunaD]

(02-06-2012 06:53 AM)Ezmode Wrote:  Thanks for the explanation, I'll make changes based on that. A couple questions, how hard would it be to readjust your spreadsheet for the sake of calculating for Annihilation, Rage, or any other Warrior tree? And, ill you be taking Vicious Throw into account at any point?

I have played Carnage/Combat to 50 (soon, twice) and a working knowledge of how the spec plays is kind of important when trying to model it. The method used (and described in my post in the Mechanics forum) is pretty general, but the real work is trying to incorporate all the various procs in a consistent way.

I suppose Vicious Throw could be included, although it would require creating two more "rotations" (sub-20%, and sub-20% with Berserk). That's something I'll plan on doing before too long, although I wouldn't expect it to make a very big difference in the weights (Vicious Throw is a melee ability which has only a main-hand attack, making it somewhere in between other melee abilities and force abilities in terms of its Accuracy dependence, and the weights are already a mixture of melee and force abilities).

I am also sort of wondering if Force Choke should be included, but thanks to the long cooldown, it probably doesn't make much difference either way.

Edit: February 10, 10pm PST
Updated with Force Choke and the 1.1.3 Surge nerf

[LagunaD]

I have been working on incorporating Vicious Throw in the spreadsheet, and seeing some surprising things. Despite the fact that it has a nice tooltip damage, it looks like Vicious Throw needs to be used selectively, if at all.

My starting assumption was that you would use Vicious Throw on cooldown, when possible. Since Vicious Throw costs the same as Massacre (when Berserk isn't active), it would more or less replace Massacres when not on c/d and not Berserk.

It's immediately obvious that Vicious Throw is never worth using when Berserk *is* active. It doesn't get a cost reduction or a GCD reduction, meaning it underperforms Massacre (plus the automatic Ataru) in both DPS and DPR while Berserk is active.

The surprise is that even when Berserk *isn't* active, using Vicious Throw on cooldown is essentially DPS-neutral. The main reason is that Vicious Throw doesn't automatically proc Ataru->Blood Frenzy OR give the 6-second buff to Ataru procs afterward. The speadsheet says that using Vicious Throw on cooldown costs you an average of almost 4 *random* Ataru procs per minute (in addition to losing the automatic procs from Massacre) due to not having the buff from Massacre up. It also costs about 2 Rage per minute lost from fewer Blood Frenzy procs, reduces the uptime of the buff from Execute, and reduces the number of opportunities to use Force Scream immediately when it comes off cooldown (since if Blood Frenzy isn't up, you need to Massacre first).

Vicious Throw is only about half an Ataru better than Massacre+auto-Ataru, so missing the Ataru buff for even one GCD is pretty significant.

Since this is a spreadsheet, and not a simulation, correlations between skills used one after the other aren't modeled well, of course. With some selectivity, Vicious Throw is almost certainly a DPS increase, but the spreadsheet strongly suggests that using it on cooldown is a bad idea.

The best time would seem to be when the Massacre buff and Blood Frenzy are already up, and not about to expire. Then you aren't losing any buffs by using Vicious Throw in place of Massacre. Conversely, when *neither* the Massacre buff nor Blood Frenzy are up, you should almost certainly use Massacre, even if Vicious Throw is available.

[Monaerch]

Quick question, sorry if it has already been answered.

I fully agree with what you are saying about vicious throw, as it really doesn't provide any synergy into the rotation. It got me thinking though, does it benefit from the massacre buff, as in can proc ataru? Since it is ranged I don't know if it counts as a melee attack?

Also would force scream and saber toss be the only other abilities that dont proc ataru?

Aaaaand would keeping the massacre buff up 100% trump using the big abilities on CD, say holding off on a battering assault or gore for 1 GCD in order to gain the massacre buff before using the ability given that you have enough rage to do both?

[LagunaD]

(02-13-2012 10:17 AM)Monaerch Wrote:  Quick question, sorry if it has already been answered.

I fully agree with what you are saying about vicious throw, as it really doesn't provide any synergy into the rotation. It got me thinking though, does it benefit from the massacre buff, as in can proc ataru? Since it is ranged I don't know if it counts as a melee attack?

Jedi do not have "ranged" attacks; only melee and force. Despite appearances, Vicious Throw/Dispatch is a melee ability. As such it would be affected by the Massacre buff and can proc Ataru randomly, but it involves only one attack, compared to 2+ attacks for most other abilities.

(02-13-2012 10:17 AM)Monaerch Wrote:  Also would force scream and saber toss be the only other abilities that dont proc ataru?

Anything listed in your ability panel as a Force ability will not proc Ataru. This includes Smash, Savage Kick, Pommel Strike, and Force Choke, in addition to Force Scream. Vicious Throw and Deadly Throw are both melee abilities, and have a chance to proc Ataru.

(02-13-2012 10:17 AM)Monaerch Wrote:  Aaaaand would keeping the massacre buff up 100% trump using the big abilities on CD, say holding off on a battering assault or gore for 1 GCD in order to gain the massacre buff before using the ability given that you have enough rage to do both?

This sort of question really requires a simulator. Having the buff up for a Gore or Battering Assault is worth at least 80 extra damage, on average. Having it up for an Assault is worth almost twice that. That seems potentially significant.

[Dunadan]

I expect that even with the surge change surge will still be the third best stat to get after strength and power?