I have a list of small changes I want to make, but I am waiting till Tuesday, so if you want to post about these, just wait till tomorrow:
- fix to damage distribution details in combat tab for advanced users (reason: several links broke when I changed the way crit cap was implemented)
- Runed Band of the Kirin Tor has conflicting stats between ptr.wowhead and mmo.db
- Add the ilvl245 bow and cosmetic changes proposed by Sp00n (that bow is actually extremely good for combat in 3.3 and the interface changes are pretty helpful, thanks Sp00n)

Dynapse: you can help by testing it yourself, I do not have time to compare hit and haste when not in danger of the crit cap.

To address many questions of the type "3.3 is coming out tomorrow, and I am planning to spec mutilate, what should I spec and gem for," I am going to share the results I have obtained using the simsheet. But before I do, I need to warn you that best results should be obtained by you directly plugging in your gear and playing with various specs and gems. These results are only valid for the gear I selected and only to the extent the beta version does not contain any major bugs. I am reasonably certain in validity of the latest file, but bugs happen.

I can further add that when switching to ICC gear and t10, optimal specs and gemming WILL change. You will run into crit cap problems. It's likely that you will have to regem, respec and change your profession to get the best out of your dps.

I have done the tests with the gear set that I believe represents a median reader of these forums:
245 level of t9
258 level of offspec items except for trinkets and bracers (I used 245 version of DV as one of the choices)

Given these restrictions, the list of gear was as follows:
Twin Spike Heroic
Steel Bladebreaker Heroic
BRK-1000
VanCleef's Helmet of Triumph
Charge of the Demon Lord Heroic
VanCleef's Pauldrons of Triumph
Drape of the Untamed Predator Heroic
VanCleef's Breastplate of Triumph
Bracers of Dark Determination
VanCleef's Gauntlets of Triumph
Belt of the Merciless Killer Heroic
Legwraps of the Broken Beast Heroic
Treads of the Icewalker Heroic
Ring of Callous Aggression
Ring of Callous Aggression Heroic
Shard of the Crystal Heart
Death's Verdict


The preliminary findings:
- I have tried many types of gemming for 51/13/7 and 51/18/2 and found 51/18/2 to be firmly ahead
- In terms of caps: reaching expertise cap is key, also in this gear you are below crit cap and above poison cap, regardless of gemming
- AP gemming is ahead of agility gemming even when you are not running into crit cap problems (significant results if you regem all your gear)
- Berserking is slightly ahead of Mongoose, but the results are not statistically significant (Mongoose will become more of a problem later due to crit capping)
- Crit and hit are worse than haste (can't tell you exactly by how much till EP values are calculated properly). Obviously if your gear differs and you are having poison cap or crit cap problems, hit rating will gain a lot of value.
- AP is better than haste, however there is no significant difference between Shard of the Crystal Heart and Mark of Supremacy. That is probably in part due to the fact that I am timing the trinket proc with Heroism, and haste rating scales better than AP with heroism.
- Engineering still comes out ahead if everything is used on cooldown and not counting the intangible effect of nitro boots. However, this is also one of the things that may change the more you move into ICC gear territory because of the crit cap. For now I am seeing about 50 dps gain compared to other alternatives.
- Quick comparison to poison-based combat shows it is about 2k below in dps, but I did not try to optimize it by trying haste and hit gemming.

I just want to add, When 3.3 comes out, the lvl 251 kirin tor ring will be better than the nonheroic Ring of Callous Aggression.

The following is the link: Runed Band of the Kirin Tor - Items - Sigrie

The gain is

+4 agility
+6 AP
+7 haste
+1 crit
+4 Stamina

Mavanas, with using Shard of the Crystal Heart as well as being an Engineer with Hyperspeeds, is there any foreseeable cap for haste?

Unlike casters, there is no real haste cap for melee classes. I suppose you can say that you cap on haste if you gain more energy than yo can spend, i.e. more than 55+55+35=145 energy in three seconds (just thinking of mutilate + mutilate + envenom spam), or 48 energy per second, 45 per second after adjusting for focused attacks from specials. That means with 50% crit rate from autoattacks, you would need to land 90 autoattacks per second (more if some miss or get dodged). It would take 7000% haste to reach that level. Currently, highest haste levels with all procs can probably reach around 540%, so we have a very long way to go. My simsheet will break down much earlier than we'll reach those levels. I will run into discreteness problems when haste levels start reaching 600%.

P.S. Shard and Hyperspeeds do not stack, you have to wait for one proc to be over before you can use the other one.

I was intrigued that you found 4pc245 worth keeping, I dropped mine in favor of Broken Beasts and Calamitious Fate, later added the Spaulders of the Snow Bandit, also Silver Assassin simmed higher than 245 triumph, but I've had no luck with getting it to drop.

For that matter 4pc t9.258 was only a minor upgrade over the alternative, so I never bothered with it.

Yeah you can change cell C28 in the Talents tab to say "=B28*10%" instead of "=B28*15%"

I am just waiting for more data on the proc items before I update.

Got a question about the 3.3b SS. I have done a lot of searches and haven't found the answer yet.
I've run the EP calculations twice now and both time it places haste at 2.22 EP value. Logically thinking that would put a 20 haste gem above 40 AP gem right? But when I switch the gems and check DPS the AP gem always comes out like 30 dps ahead on about 4 consecutive 1000 iteration runs.

Is this due to a problem in how EP values are being calculated? What should I follow, the DPS from the sim or the EP value from the EP test?

On a side note. You cannot search for EP value because the search feature doesn't allow anything less than 3 letters and there wasn't much to find with equivalence.

Are you looking at the right row? Is that the EP value of haste or it's dps per point. Because as far as I remember, AP was at about 1.1 dps per point in 264 gearsets which would make it about equal to haste.

I put in my current gear which is 4/5 of 9.245 and mostly 245 in the other slots. I'm positive I looked at the right row.
Row 10015, Column G of DPS output page. Average EP.

If you are absolutely sure you did not confuse EP and dps/point, then the only answer that I can think of is RNG. Typical standard error for EP values is about 0.07, so if you did see the EP of 2.22 I can tell you with 95% confidence that haste is ahead of AP in your gear setup.

On the other hand it is theoretically possible (although in my mind very unlikely) that unlike the EP result, in a thousand iterations AP comes out ahead 3 times in a row. So check the p-value in the inputs page, is it less than 0.05? If it is very small, then the result is statistically significant and something just does not add up, and it can't be the EP calculation because it is such an easy macro and I've been using it forever. If the p-value is more than 5% it leaves some room for what you found to happen just due to inherent fight RNG.

I did EP calculations for new gear before and haste actually came up at 2.28 while AP was 1.1 but that was dps per point, not EP, which puts haste a hair ahead of AP. So that's why that was my guess that you somehow confused the EP values and the dps/point since the order of the two units is so similar. Maybe if you saved your EP calculation, tell me what was the value you got for each stat, including AP.

If nothing works, you can send me your sheet or put a link to it. I'll check myself since what you are reporting sounds very dubious.

I ran the EP test again with the new change to HFB. My AP has EP Value 1.0, Haste has EP value 2.34
Switching my yellow gems out, -100 AP +50 haste ended up being lower dps in 3 runs of 1000 iteration, HOWEVER the p-value box was grey showing that the change was NOT significant. I also tried swapping AP gems for the haste+AP gems and the change was NOT significant.

Yeah the difference in the dps from switching those gems was not big enough to be significant at 5%, at least with 1000 iterations. Now you can look at your EP values, and below them you will see a row called SE. If 1.96 times SE is smaller than 0.34 then you can say with 95% confidence that haste is better than AP for you.

Ragarding your question about EP calculation, it is done the same way as any regular simulation, just 10,000 times for each stat. Nothing changes in terms of buffs or abilities you use. Basically the macro sets iterations to 10,000 and executes the simulation code. Then repeats for each stat.

has anyone tried using the enchant Black Magic? My understanding is that you gain nothing from having it on more than one weapon so a combination of zerk\black magic would be the way to go for a test. Or perhaps you MH the enchant, not sure.

No black magic is weaker than both berserking and mongoose, you don't need a simulation to figure that out. However, it is worthwhile to have it on an extra ofhand and swap on for a couple seconds for the proc, then back to your regular berserking offhand. You should do it every 35 seconds, and it's worth to have some addon like proculas to see when black magic internal cooldown runs out. You can do the swapping easily without any addon just using a macro.

By swapping BM offhand you should be gaining around 80 dps, but I am adding that to next version of the simsheet that's currently in the works so that you know for sure.

I dont know if someone mention it before but i think it would be nice if u could add like "Socket 1 (Head)" to the socket section in the sheet. So it says what part the sockets on.

And thanks for a awesome sheet.

Do you have any thoughts on Heartpierce, as I did not see it in the 3.3b Sheet? Also Whispering Fanged Skull has a red box around it.. does that mean there's no data for it?

Last week I upgraded to several ICC pieces, lost a lot of haste, but gained a ton of crit. With the new gear, I want to be sure to maximize my gemming, and was curious to use the sheet to see what impact gemming 3x Haste Dragonseyes instead of AP would do.

Also I was wondering if you had an ICC BiS list for guidance.

I have a 3.3c version I am working on and it has Heartpierce, WFS and Deathbringer's Will among other things. Just be patient, I am trying verify the proc data before I release the new version.

As for EP values, the differences you are observing are bigger than what can be explained with rng alone, but unfortunately, it's hard to determine without a careful analysis. Both files can still contain bugs, but theoretically, EP for ARP can be lower if:
a. Incorrect formulas were used to determine physical damage reduction
b. Incorrect uptime was used for any arp trinkets
c. Incorrect formulas were used for any attacks that depend on armor penetration, i.e. autoattacks and mutilate.

To find the source of the error, I'd start with determining the dps and portion of damage that comes from autoattacks and mutilate. Before I dig into all that, wait for a new version of both sheets and compare the EP values then.

For 51/18/2 build the spreadsheet shows a significant benefit from choosing relentless strikes over opportunity. I'm curious as to the assumptions in the caclculations of these two talents. Is the returned energy factored through the simsheet at the (randomized) time of the process an then the energy applied as best case scenario staying consistent with mutilate to 4CP them envenom or us it calculate upfront base on the expected number of procs etc. ?

The reason I ask is that mentally thinking about it an reviewing data from 25m raids I have been on 20% bonus dmg to mutilate still comes out way ahead compared to the energy gain from relentless strikes alone. I'm sure I'm completely off somewhere here considering I don't always run with every single 25m raid buff.

No it's a simulation hence events occur in the simsheet at the same time they would occur in the real combat. So each time you do an envenom, the simsheet rolls a dice to determine if relentless procs or not. The probability of relentless proc is the same as your talent points suggest. Implementation of opportunity is equally straightforward, i.e. mutilate damage is increased by the factor suggested by your talent points.

Simulation is a fairly straightforward programming approach, the only trick when implementing a simulation is to write a priority of actions that would properly represent your decision making while in actual combat. Which bars and indicators are you looking at in a raid? You track energy, combo points, snd buff, envenom buff, rupture buff, various talent cooldowns, like AR, BF, vanish, KS, DP stacks, and trinket procs. Out of all these indicators, the simsheet uses everything except for the trinket procs. And the reason I did not write an AI to be based on trinket procs is not because it is hard to program, but because it is hard to execute in a real fight and trying to do so makes you lose track of other important timers. I am trying to keeping the simsheet logic simple and realistic.

This idea: "Is the returned energy factored through the simsheet at the (randomized) time of the process an then the energy applied as best case scenario " describes an unrealistic scenario that appriori biases the result, so I would never consciously program that in. I am not trying to scew the results in any way.

As a side note, there is plenty of anectodal evidence that 51/18/2 pulls and it is also confirmed by Aldriana's calcsheet analysis. Finally the difference between the two specs is so close in the simsheet, that is not at all unlikely to find that by RNG 51/13/7 would win. I posted these results elsewhere in these forums, but I can repeat them here. I ran 51/18/2 and 51/13/7 side by side and then did a pairwise comparison of the results, as if two rogues with those specs competed in a single traget fight. In 156 out of 300 iterations, 51/18/2 won, but still there were 144 iterations where 51/13/7 ended up ahead. That's how close the two specs are in terms of their dps.

I am releasing a new version. Apart from a couple minor bug-fixes and preset changes, I have added the following to the sheet:
- more loot and more procs implemented. Out of the procs, Black Bruise has a tentaive implementation and Tiny Abomination in a Jar does not have an implementation yet. Everything else should be working properly.
- HfB has been reduced to 10%
- Changed DPS rules slightly to manage envenom buff better and to allow users to implement hybrid cycles, such as 3-4+ discussed in the mutilate thread
- for gear selection, I ran ilvl264 EP values for combat and mutilate and added an option to value capped stats as likely alternatives, such as valuing expertise as attack power and hit as haste
- added hysteria

I also tried Black Magic, which I expected to perform well, but closer look at the losses revealed it is not worth swapping it throughout the fight. I also did a preliminary analysis of BiS for ilvl264 for each spec and marked the gear in the Gear tab.

Mutilate:
- AP and Haste gear while staying above the expertise and spell hit cap and below the crit cap
- 51/18/2 spec with RS
- berserking enchants
- engineering as one of the professions if using bombs on cooldown
- Heartpierce modeling has changed based on recent findings. It appears to proc from IP procs as well as DP refreshes. With these changes Heartpierce is the best mainhand for ilvl264.
- Deathbringer's Will is a poor trinket for mutilate. It appears that Whispering Fanged Skull Heroic and Herkuml War Token are the best trinkets at level 264, but of course Abom can change that. DV heroic is still really powerful.

Combat:
- ARP and Haste gear while staying above the expertise and spell hit cap and below the hard crit cap
- 20/51/0 spec (I spent very little time ironing out the distinctions between VP and IP as well as Imp Evisc and Ruhlessness)
- berserking enchants
- engineering
- Black Bruise in it's likely implementation does not work as well as Hack and Slash
- Deathbringer's Will works reasonably well and the second trinket appears to be Whispering Fanged Skull Heroic , however Herkuml is not far behind
- trying to stay below the soft crit cap as well as the soft armor pen cap is not the best strategy. The uptime of going over those caps with DW procs is jst above 9% so it's not worth adjusting your gems for it. Basically DW is a good trinket, but some of its proc value is always wasted.

Combat is about 1600 dps below mutilate (15%) on a single target.

I am sure there will be a lot more questions for which I do not have answers yet. DW trinket has complicated analysis a lot in terms of finding optimal gear, so it will be a lot of trial and error. It is possible that AP/haste combat alternatives perform very close to ARP/haste ones. Mutilate is a lot more straightforward but seeing how close 51/13/7 is to 51/18/2, it is possible some other setups can emerge. So try it for yourself. The new sheet is located here.

- nm - disregard