Rogue DPS Simulation Spreadsheet - Page 25

AzGuL^ · 10/02/09, 12:02 AM

A guldie told me (after I ran WP/WP vs WP/DP through the spreadsheet and WP/WP came on top with about 40 DPS) that there's something wrong with this spreadsheet and I should be using Aldriana's. Anyone know why he is of this opinion? I was just told 'I just know it's better'. Is there some truth to it?

I was going to run the same thing through Aldriana's but I was not able to find the option to change OH poison.

Mavanas · 10/02/09, 1:05 AM

I am going to speak for myself, but I am certain a lot of what I am saying also applies to Aldriana's work. The spreadsheet you are using represents my best effort to model rogue dps based on simulation approach. I have cross-checked its results against Vulajin's and Aldriana's spreadsheet at various points of its existence, and found errors in all three. No spreadsheet is safe from errors, but I personally do my best to catch them during testing phase based on inputs of various users.
Also you should understand that this spreadsheet differs drastically in its approach to modeling dps since it's a simulation. There are pros and cons of both approaches. I personally consider simulation a better approach to match actual fight environment and to model various synergies and procs, and therefore I consider it more realistic. At the same time, it has assumptions of its own and it is still a model, so it's only accurate to the extent it correctly represents your behavior in combat.
Aldriana definitely has a lot more experience modeling rogue dps and his work is widely recognized. Nonetheless, as a biased programmer, I trust my work more in terms of its accuracy and I just wish there were a way to minimize the variance of its DPS estimates. This said, you will find that the conclusions of both spreadsheets have been very similar, and that's a good sign. It means we are both doing things right to an extent (or making same mistakes ^^). So for instance, I believe both spreadsheets suggest using double WP for combat and moving toward ruptureless cycles. At the same time, I have not seen Aldriana's model of swapping offhands, and I am not even sure he's going to attempt to make one, so I cannot compare the most recent findings, such as gemming AP, and enchanting mongoose in 3.2.2 BiS gear while weapon swapping.

• Aldriana · 10/02/09, 1:40 AM

To touch on a few points briefly raised by Mavanas:

1) I have not implemented weapon swapping, nor do I plan to. So if you want to look at the implications of that, Mavanas is the only game in town. Either a) it's not a DPS increase at all, or b) it is, so it will be nerfed, likely in 3.3. Either way, I don't feel like spending time on it.
2) Like Mavanas, I happen to believe that my modeling is superior for the usual sorts of questions you might want to answer - partly because I believe Simulators to be an inferior tool for certain types of optimization, and partly for vaguely egotistical reasons based on the fact that I do have more experience with such matters. Many rogue mechanics have subtle or counterintuitive aspects to them, which inevitability get modeled incorrectly at times - a good rule of thumb is that all spreadsheets contain bugs. Since I have been working with rogue theorycraft for longer than most people here, I happen to believe I do a better job of avoiding such bugs, which in turn leads to greater (if still imperfect) accuracy.

To be clear: I have no specific knowledge of bugs in Mavanas's sheet, and I'm not trying to criticize or demean his efforts - it's simply that in my experience, there are just too many things that are too easy to screw up, and I've yet to find a spreadsheet that doesn't contain any. When I tore apart Combat 1.2 to write Combat 1.3, I found bugs. When I tore apart the old Mutilate sheet to create the new beta, I found bugs. There are bugs in the old Rogue Gear Sheet v0.10.4. I have found bugs in Vulajin's spreadsheet. I have found bugs in the old Rogue DPS spreadsheet. As such, I think the safest conclusion is that every spreadsheet contains bugs, and accuracy is thus at least in part a function of one's ability to minimize the number. And perhaps this is egotistical, but I'm of the opinion that the extra years of experience give me an advantage in this regard.

As such: I happen to think my work is more accurate than Mavanas's. As Mavanas just stated, he feels similarly. As such, the one that you should trust is a function of who you happen to trust more, which you find easier to use, and so forth. And the only person who can decide that is you.

Axodry · 10/02/09, 1:19 PM

Hi Mavannas,

Thanks for putting this simulation together. Do you have plans of simulating the effects of fleet footed into a Mut/Envenom dps strat? I know that the current 51/18/2 Mut/Evis strat can be changed such that the finisher is envenom instead of Eviscerate, but i'm not sure how to show the benefit of fleet footed. I made the changes and i'm coming out with slightly less than a 1% dps loss, but again, i'm not showing any benefit of having increased movement speed (also, i'm nowhere near BIS as a sidenote..i'm using 2 t8/2t9 since with this spec).

Thanks again.

Mavanas · 10/02/09, 1:27 PM

I do not have any plans to model Fleet Footed because it would complicate the spreadsheet significantly. At the same time, I've seen several napkin math analyses on these forums of run speed effects. Just search for them.

Mavanas · 10/03/09, 2:03 PM

Mutilate

Assumptions: Best 3.2.2 gear, with weapon swaping, using same offhand for both DP and IP
Spec: 51/18/2
Professions: JC/BS (LW, enchanting, or alchemy)
Gear:
Steel Bladebreaker MH (mongoose, IP)
Steel Bladebreaker OH (mongoose, DP/IP swapping)
BRK-1000
VanCleef's Helmet of Triumph
Charge of the Demon Lord
VanCleef's Pauldrons of Triumph
Vereesa's Dexterity
VanCleef's Breastplate of Triumph
Armbands of Dark Determination
VanCleef's Gauntlets of Triumph
Belt of the Merciless Killer
Legwraps of the Broken Beast
Treads of the Icewalker
Planestalker Signet
Ring of Callous Aggression
Comet's Trail
Death's Choice

Enchants and gemming: Standard enchants, six 20AP/10haste gems, three 68AP gems JC-only, 1 prismatic gem, rest are 40AP gems. AP food.
Rotation: Envenom-only (4+), 60 energy pooling, no extra priority if envenom buff drops. Swap offhands with 3 seconds remaining.

----------------------------------------------------------------------------------------------------

Combat

Assumptions: Best 3.2.2 gear, with weapon swaping, using same offhand for both DP and WP
Spec: 18/51/2
Professions: JC/BS (LW, enchanting, or alchemy)
Gear:
Stormpike Cleaver (berserking, WP)
Lionhead Slasher (berserking, DP/WP swapping)
(the rest of the gear is the same as for mutilate)

Enchants and gemming: Standard enchants, six 20AP/10haste gems, three 34haste gems JC-only, 1 prismatic gem, seven 20hit gems, the rest are 20 haste gems. Haste food.
Rotation: 3/5/5 with eviscerate and rupture, 60 energy pooling, refresh SND 2 seconds before expiration or 5 seconds early if it is due to expire at the same time as rupture, pause eviscerate if rupture is due in 2 seconds. Swap offhands with 3 seconds remaining.

Aéquitas · 10/04/09, 8:28 AM

In the light of weaponswapping what would be the ideal time to switch from deadly poisons to instant poisons on a fight like faction champions?

My gut feeling says that changing the script to swap to instant poison as soon as you got 1 stack of deadly poison on the target so you can use envenom as a finisher would be the best. The "bosses" die within 30 secs most of the time so building 5 stacks might not be ideal.
Is this assumption correct?

Pandeimos · 10/04/09, 9:19 AM

Intresting results Mavanas, especially given this discussions on combat with the combination of soft armor pene cap & evis only cycles. Seeing you have a lot experience with these simulations how do you think the weapon swapping aspect influences these results? I.e. would the results be the same for none weapon swapping (considering blizz does not want us to use this), would we see completly different gear setup?

Mavanas · 10/04/09, 10:15 AM

Yes it's slightly different without weapon swapping. For mutilate best spec is 51/13/7, gemming for agility, still envenom-only cycles with double mongoose and Beast legs offpiece. For combat without weapon swapping, it's 15/51/5 spec with soft armor pen cap, with double wound, double mongoose, and Calamitous Fate chesptpiece as offpiece.

Lukon · 10/04/09, 11:13 AM

Originally Posted by Mavanas

For mutilate, in 3.2.2 BiS, with weapon swapping EP values are:

AP	1,00
ag	1,97
hit	1,72
haste	1,98
apen	1,36
crit	1,82
exp	1,92

Standard error of the the estimates is 0.06. So as I predicted from previous tests, AG and haste are just below AP.

Could you explain how you derived these values, Mavanas? From what I see in the DPS Output sheet, it looks like you add 50 to each stat and use 1000 iterations - is that correct?

rENaitre · 10/04/09, 11:19 AM

I've been experimenting with wep swapping since it popped up on the forums here and hands down, it's amazing. On longer fights where I run a high Rupture cycle with 51/13/7 or short fights 51/18/2, simply swapping out has made me able to compete with much higher geared Combat Rogues. I'm not very close to 3.22 BiS so mostly I raid in Rupture spec.

I'm an Engineer and was surprised to see that Hyperspeed Accelerators weren't mentioned as being beneficial. I'm also wondering if Shard of the Crystal Heart might be something to pick up until the BiS trinks.

Mavanas · 10/04/09, 12:05 PM

I subtract 50 of each stat and then use 10,000 iterations to calculate EP values.

As for hyperspeed accelerators, they do in fact look very good with weapon swapping. 340 haste for 12 seconds every 60 seconds, is 68 haste, which in BiS gear is 134.64 EP, so the benefit of the profession is 90.64, just from the gloves. For comparison, BS and JC are both 80 EP. If you also consider that nitro boots are slightly better than icewalker and speed boost is good on some fights, plus 27 dps from bombs used on cooldown, so engineering really shines. As for the shard, it's still not worth it since 512 haste is still worse than 1024 AP you can get form Mark of Supremacy.

Lukon · 10/05/09, 12:47 AM

Is it meaningful to give EP estimates to 2 decimal places with a standard error of 0.06?

Starting with the original sampled variables for 2AP (mean = 2, standard error = 0.06) and Agi (mean = 1.97, standard error = 0.06), we can create a new normal variable (2AP - Agi), which will have mean 2 - 1.97 = 0.03 and standard error sqrt(0.06^2 + 0.06^2) = 0.085. Integrating from negative infinity to 0 gives a value of around 0.36. ie there is a 36% chance that agility is better value than attack power.

Mavanas · 10/05/09, 1:48 AM

It's normal to round point estimates to the number of significant digits in the standard error.

I did not check your probability calculation, just guessed you assumed normal approximation, but it does not strike me as out of ordinary. The EP estimates I am getting serve as an indication and can be used to calculate EP values of various gear pieces. Items with close EP values still have to be run though the simulatior for a final comparison. If your DPS with one item is not significantly different in statistical sense than the DPS with some other item, chances are you won't notice a difference if you did same fight for a year day in and day out, so go with the one that gives you a higher average dps. When I reach a conclusion about gemming AP or AG, it's done before the EP calculation. I replace several gems and observe which dps is greater and whether the results are statistically significant.

In the end, these EP values are still my best estimates given the number of iterations and more often than not they confirm my prior knowledge about ordering of stats.

SilentNSly · 10/05/09, 5:35 AM

Originally Posted by Mavanas

Yes it's slightly different without weapon swapping. For mutilate best spec is 51/13/7, gemming for agility, still envenom-only cycles with double mongoose and Beast legs offpiece. For combat without weapon swapping, it's 15/51/5 spec with soft armor pen cap, with double wound, double mongoose, and Calamitous Fate chesptpiece as offpiece.

Thanks for mentioning non-weapon swapping too. I suspect weapon swapping will get nerfed very soon (especially since Ghostcrawler said they will take action if it was better)

1) I am a bit confused of "with soft armor pen cap" as there are no trinkets that proc armor pen in your BiS list.

2) Does this mean that you would replace the gems with armor pen gems?