Actually by the time you build up the CP's to put up SnD wouldn't it be more beneficial to save the CP's, pop KS & BF during MR proc and then pop SnD, AR, etc ?

Surely you have time to build SND up before MR proc runs out and you have to KS. All you need is 2 sinister strikes + snd, that's 3 global cooldowns. Besides you have to burn energy before a KS anyway, so buiding up SND is a good way to do it. In fact if you want to go into that much more detail, the best strategy might even be to start with garrote, SS, SND - that's three seconds, then continue with SS to keep energy low while watching your MR proc. If it's a late proc, you should be able to apply a rupture as well. If it's early you will only have time for snd before you have to use KS.

Btw another interesting case is XT where you want to maximize DPS in 30 seconds using cooldowns, but you have the luxury of having full combo points before the heart phase so you can start with SND running already. The strategy there depends on whether you stopped dps long in advance so that your trinkets are off cooldown as soon as you attack the heart. If that's the case, then you burn your energy (2 sinister strikes), apply rupture, BF, and KS right after. By then your trinkets will most likely have procced and the heart has 3-4 sunders already. Then follow up with AR.

However, if you did not stop dps and your trinkets are on cooldown at the beginning of the heart phase, it's actually better reverse AR and KS. So burn energy, use AR, BF,and then KS closer to the end of the heart phase.

I think one important thing speaks for 3/3 MP imho, besides the 10 man crit buff, if you have no Retribution Paladin:

The Envenom buff in fights, where you swith targets frequently: For example Freya (the small lashers or the elementals) - when you need to switch the target and use Envenom before that - the Envenom buff is highly useful to have enough poison stacks for a 5CP Envenom and it's less likely, your SnD will drop, when those low hp adds are killed quickly.

Those situations aren't covered in the spreadsheet but occur quite frequently even in boss encounters. Therefore I'd say, that even a slight Spreadsheet DPS increase when going 2/1 (with my gear ~0,23% DPS increase) might not be worth the tradeoff in those encounters.

There was a discussion in Simple Q&A. I want to expand on it here because it went beyond the scope of that thread. Basic idea is to have deadly poison on, then as soon as DP is refreshed, swap to the same offhand weapon with wound poison on it, and before DP expires, swap back in time for DP to tick. Aldriana's analysis showed that the strategy is basically not worth the trouble. However, based on my analysis and I think I can see a potential in it during heroism.

My damage numbers are slightly different from those of Aldriana because he was using his gear for analysis, and I am basing this on BiS combat gear. However, there are other differences. First Aldriana claimed that refreshing DP on average gives 10.5 seconds before it has to be refreshed again not to incur any DP poison damage loss. Based on my own testing which I included in the source file, after a DP refresh, you get 12 seconds. If DP is refreshed any time before the end of the 12 second period, DP stacks are not lost, DP ticks are not reset and DP damage goes on exactly as if no offhand weapon swap ever happened.

Second, my estimate of damage lost if DP drops is adjusted for heroism. Also for the case when there is no heroism, I find it cheaper to refresh DP with a Shiv.

So here are the numbers: WP damage = 640, DP 1 stack dmage = 1127, MH damage = 2227, OH damage = 866, SS dmage = 3743, Shiv damage = 682, percent haste from gear = 11.4%, dodge = 0.9%, white miss = 12.4%, offhand base speed = 1.4, offhand speed during heroism = 0.51, swapping/reaction time = 1 second.

With these inputs, highest DPS increase is found by using wound poison for 9 seconds. Due to reaction time of 1 second, you will have 2 second remaining after swapping DP dagger back on. There is a 29.6% chance DP will not be refreshed before it runs out at these levels of haste, dodge and miss rates.

Damage gain from wound poison is 9*640*50%/0.51*(1-0.9%-12.4%)=4916. As far as cost is concerned, there is cost from lost white damage swings of 2227+866=3093. Also there is a 29.6% chance to drop DP, and when that happens, two things occur. DP ticks are delayed (potentially causing a loss of DP damage if DP does not tick an extra time before the fight is over) and there could be several ticks at stacks less than 5, which also causes DP damage loss. The exact calculations of these two components are in my backup file, but the first loss is 1378, while the second one is 1108, for a total of 2487. With heroism up, this cost is lower than the other alternative of using shiv to refresh DP poison if it does not refresh by natural causes (which I estimate as 2602 - energy-adjusted sinister strike damage).

So altogether, there is a 1088 damage gain. After DP is put back on, it will refresh on average after 1.5 seconds, so the whole manipulation with swapping takes 10.5 seconds, and gives you a 103 dps increase. Swapping can be repeated as soon as DP is refreshed again, so it can potentially yield up to 4352 damage over the duration of heroism. In a five minute fight it is a 14.5 dps increase, for shorter burst fights, it will be a much bigger dps boost. There are other times when I expect this strategy to yield a dps increase, such as on use of a speed potion and during a Comet's Trail proc (perhaps Dark matter because it only affects wound poison damage). Going forward with better levels of haste and crit, the strategy will only become more profitable.

Without those procs, the strategy is not worth it, even though you can do better by using Shiv to refresh DP.

So, the testing you have in your sheet regarding poison refreshes isn't enough to refute my characterization of how deadly poison works. My assertion was that the poison is not refreshable after the 4th tick, as exhibited by testing in this post. The relevant section of the logs is this:

Basically, there's a preexisting Deadly Poison, which ticks at 30.187. Just before the next tick, we refresh at 32.640. Thereafter, we get 4 ticks, and immediately after the 4th one, the Deadly Poison debuff drops at 42.328. Even though it's been less than 10 seconds since the last refresh, the deadly poison debuff drops, and the stack is gone.

However, despite the fact that the evidence you present is insufficient to prove it, you are nevertheless right - it appears poison mechanics have been changed since that test was performed. Here's the analogous test done today:

Again, we have an existing stack which ticks at 45.358. We refresh it (twice) before the next tick, giving us 4 ticks at 49.076, 52.180, 55.044, and 58.055. Now, by the previous log, it should drop at this point, preventing refresh; however, it doesn't. Instead, almost a full second later at 58.963, we refresh it, and proceed to get 4 more ticks - each doing the same damage as the previous ticks, so we know it's a refresh and not a new stack (as we still are doing 5-stack damage after only 1 refresh).

What is worth noting, however, is that when we refresh after the last tick, it changes the tick timing - the next tick is at 1.998, roughly 3 seconds after the refresh, but 4 seconds since the last tick. Thus, a late refresh of this sort does cost some DP damage due to this DP "downtime", but it's admittedly a lot smaller than having the stack drop entirely.

I was able to replicate your experiment and adjusted calculations for the probability of reseting DP tick even if DP is refreshed in time.

So the new numbers are 87 dps during heroism and optimum strategy is to swap DP back on 2 seconds before it falls off.
It's almost not worth it with Comet's Trail - 14 dps increase (taking into account that DP restacking will be done at regular haste level not during the proc). And it's not worth doing during Dark Matter proc at all.

Taking into account that letting DP fall off will delay replication of the strategy and thus lower its effectiveness, an alternative is to commit to refreshing DP with a shiv if it is not refreshed normally in time. With this strategy dps increase while heroism is active is 66. In a burst 1 minute fight dps gain from this strategy is 44.

I updated the backup file.

I haven't looked at the calculations, but logic would dictate that;

It's the probability of the difference between your Latest DPS and your Baseline DPS being due to more random chance. So yes, p value, gauss, null hypothesis and all that jazz.

The more iterations you run, the more confident you are that even a small difference is actual due to the change in gear and not the RNG. What you decide to set as your threshold for considering something "significant" is entirely up to you, but a good baseline typically used in statistics is a p = <0.05

A p value of 1 (100%) is essentially the worst confidence you can have in refuting the null. A p value of 0 is the best. The threshold for a "significant" difference is arbitrary. A p <0.33 is consistent with something further than 1 SD from the mean. A p value of <0.05 further than 2SD, and <0.003 further than 3 SD. Those are good generic "thresholds" for significance, depending on the work you're doing.

I just want to confirm that your understanding is correct, and I could not have done a better job at explaining statistical significance.

For practical purposes, as a simulation is running and performing its iterations, when p values start getting low (the coloring of the cell will become brighter to give you a visual tool to spot it faster), you become more sure that the differences in dps between the baseline case and the latest case could not have happened by mere chance as a result of RNG, but rather are attributable to whatever changes you made to gear, rotation, spec, etc.

When I edit my current gear with Insanity's Grasp, Golem-Shard Sticker and Relic Hunter's Cord I gain about 100 DPS. If I do the same in Alriana's sheet I lose 20'ish DPS.

I already found out from Aldriana the Rune-Etched is amazing for a 163 dps weapon, but this difference would indicate the above fist/dagger setup does beat it?

New version is out. Lots of new gear and 3.2 updates. If you see any data entry errors, please let me know. All tier gear is in, however 2-piece bonus is NOT working yet as I do not know the proc chance. If you have done testing of the proc, let me know.

Biggest changes concern HAT spec. Because the spec was changed significantly, treat the DPS Strat and talent presets with caution. You have to find a strategy that works best for your own group and gear. Since new cps rates are not known yet for every class, use four copies of the highest available in the list as an approximation. In actuality I suspect the actual cps rates can be much higher, so check with the HaT thread for updates.

I was thinking about Blizz idea to give different names to gear for Horde/Alliance.
Is there any chance for implementation of some kind of IF function which will display gear name according to your Race choice? I'm not much into Excel and have no idea how to do it but it's something worth to see imo.

edit:
I came up with something like this:

I will add those items and a few others I missed in the next version. I'll try to have it out tomorrow. As for uptime calculation, the numbers on the front page are just for your information, they don't really go in dps estimation. They are accurate on average, but as far as actual uptime works, for every fight it will be slightly different depending on the rng. Trinket procs are completely done using random number generation to simulate procs chances and cooldowns.

Avarkin your idea is not bad except it would mess with the way I implemented gems right now, so I will think if an easy work around is possible.

Also if anyone finds out proc rates for the tier 9 two-piece bonus, do share! I'd love to implement that bonus.

I have added more items to the gear list, including all the new trinkets I could find. You will also notice that some of the old Naxx items are gone from the list. The new version is now available for download on myotherdrive.

I also did some preliminary analysis of the new trinkets using 3.1 BiS gear. The results in 3.2 BiS gear could be different, in particular, I would expect the combinations involving MR to drop down in value, due to higher values of Arpen from gear. I did the analysis for combat and HaT only for now. Here is the legend for new trinkets: C25 = Coliseum 25 Badge DPS Trinket, VC = Victor's Call Heroic, BV = Banner of Victory, DC = Death's Choice.

Eventually I suspect CT+DC or BV+DC to pull ahead as we have more Arpen from gear.

I have a question in relation to the 10/18/43 spec. I ran this tonight in a fairly stacked group and had nonstop evis spam other than refreshing slice and rupture. Once I realized this, i switched my MH to Fang of Oblivion from The Masticator. However, the simulation has a considerable DPS loss when equipping the dagger. I upped the group crit rate to 4/second (which is probably too high) and removed the use of hemo since I never used it except once at the start. I assumed that 188 dps > 178 dps as evis ignores weapon damage, and I don't see the stats being 150 dps apart. Poison proc mechanic? Something else?

The dps loss is entirely from wound poison, a 1.8 speed weapon has a much lower proc chance on each of those eviscerates than a 2.7 speed. Since you're literally spamming an instant attack that can proc poison every GCD, the difference in weapon speeds is amplified.

Mavanas, thanks for all your hard work on this sheet!

I just wanted to let you know that the Lionhead Slasher (Alliance Axe OH from ToC 25) has, in fact, had its speed dropped to 1.5 to match the Horde verion.

Edit: Just to clarify, at this time the Armory is still showing it as 1.6

I never personally checked it. If you were to try it with the simsheet, I'd load a lot of armor pen with MR, pick up improved eviscerate, make sure you have turn the tables instead of MP.

I have not tried all combinations, but so far I see about 130 dps decrease from such a switch, but you can try more combinations on your own.

Mav,

I downloaded your version of the spreadsheet, but when I open it(Excel 2007) I get this error in the bottum right..

"Excel was able to open the file by repairing or removing the unreadable content"
"This workbook has lost its VBA project, ActiveX controls and any other programmability-related features"

Could anyone help me here?

I think it means the download was corrupted and when Excel repaired the file, VBA programs got lost. The file will not work without the VBA code, so you need to redownload. If it keeps happening, let me know and I will put it on filefront as well.

i have some problem with spreadsheet , when i open it i see ~9xxx dps , when i turn item or weapon from 1 to null and then swith back i see difrent dps , i try this few times and have same results, dps change every time. Also when put my items and save exel file i see dif dps on every open it

Another q did Fury axe C25 normal have so low dps that you put only hero ver of it?

You are looking at a DPS snapshot, which means a single realization of your dps if you were to fight a boss once. Whenever you change something in the sheet, it recomputes all cells and gives you another estimate of your dps. Just like in real combat if you were to fight same boss, your dps would be different every time.

However, if you were to repeat the fight 300 times, the average dps over those fights would be a somewhat stable number, i.e. the effect of RNG diminishes if you repeat the experiment. To get this more reliable estimate of your dps in the spreadsheet, you have to go to tab Inputs and Run DPS Simulation. That's the only reliable way of using the simulator, the dps snapshot on the front page is "highly subjected to RNG" as it says right next to the DPS number. I will remove the snapshot in the next version since it confuses people more than helps.

As for normal version of some weapons, I tried to include all, but if I missed something, I can add it next time.

Have you considered putting a way to run the DPS simulation on the gear page? to make it more accessible to people who have more experience with more traditional spreadsheets.