I like the idea, unfortunately my knowledge of statistics simply isn't what it used to be.

Perhaps there would be value in simulating just the difficult-to-predict mechanics (like mutilate) to try to devise formulas to plug back into the spreadsheet?

Even Mutilate I think is better done statistically than via simulation. I've actually worked out about 75% of the behavior by hand before; it's just never been worth taking the time to include that information in a spreadsheet.

In terms of writing a computer program to do this statistically: I'd be willing to write the actual computation package for it myself, but there's a lot of code around that that would need to be written as well. I mean, here's an overview of what would be involved in the ideal rogue calculator:

1) UI for entering gear and talents. Ideally this would also include an Armory fetch so you can point it at your Armory Profile and it automatically grabs your gear and spec. Would also have a full talent calculator page where you can adjust your spec, and a way of changing out gear (probably something akin to the old CTProfiles).

2) Total up all stats, procs, etc. etc. from gear, and pass that information into the damage calculator. Damage calculator chews on it a bit and spits out a damage estimate.

3) For each stat (or maybe each piece of gear, or perhaps both) you rerun the damage calculation for your gear with one additional point in that stat. You use this to score each stat; then, you run these scores against an item database to get a rough score for each piece of gear to build a list of "these pieces of gear are good".

4) Then, for each of those pieces of gear, you run the damage calculator again, to get the exact upgrade from using that piece of gear versus what you're wearing. You then use this to build a list of "if you swap out this one piece of gear, this is the benefit you'd expect to get.

5) You display a list for each slot consisting of "this is how good the item is in an abstract sense, and this is how good it would be for you" - i.e., if you're hit-capped, items with high hit would still show as "good" but not an upgrade for you.

6) Ideally, you also add zone filters and the like to restrict the list of gear it returns, and other such nice features. Note also that socketing adds a lot of complexity to the above.

So, you'll note that this contains a lot of sorting and UI-type work above and beyond the actual calculator; the damage calculator itself, while perhaps the more technically sophisticated piece that requires the most knowledge, is probably less than half of the actual code that needs to be written. The rest doesn't require any particular patience or even that much coding sophistication; just lots and lots of time and effort.

Note also that you can get away without a lot of the fancy UI tweaks and gismos, but you fundamentally do need:

a) A way of entering your gear
b) A way of entering your spec
and
c) A way of displaying a damage number and an upgrades list

even if you do nothing else - and that's still work.

So, yeah. You probably only need one programmer with actual understanding of the mechanics and statistics that go into the DPS calculations - and I'd be willing to be that one - but you'd still need some people to invest the time and effort to write all the rest as well.

What language do you primarily work in, Ald, and what parameters would you require (obviously a string-like representation of the build, but can you just sum up the gear and provide the numbers?)... I'm basically asking what the "Calculate DPS" function's parameters would be.

Well, I could probably write this in any reasonable programming language (no Befunge or Intercal, thanks) - but were it up to me, I'd probably use Python. In particular, I've been doing quite a bit of stuff with Django (a python-based web framework) at work lately, so the natural way of approaching this problem - at least for me - would be to set it up as a webapp written in Django with the underlying computation done in Python. I'm open to suggestions, though.

In terms of actual implementation details - we should probably adjourn to private messages (or, if enough people get interested, make a thread for it rather than taking over this one), but, under the above design philosophy of doing this with Django and Python, the natural inputs would probably be dictionaries, one for gear and one for spec. And I suppose at some point you'd also need to track race, buffs/debuffs, opponent armor values, and a few other such things. Some design work would need to go into figuring out what gets tracked where if this were actually going to be made to happen at some point.

Oh, and as long as we're dreaming: in a totally ideal world, the UI and general framework would want to be written in a class-independant way, such that if, sometime down the road, people were feeling ambitious, they could write a damage calculator and a talent page and without changing the underlying framework generate a similar calculator for another class.

In short: there's a fair amount of thought and design that would need to go into this if one wanted to truly do it right. I've thought through some of the issues, but there's plenty I haven't; if this project has the support to actually get off the ground, we'd need to talk over these various issues and work them out before we started coding in a serious way.

Not sure if this would be helpful to you, but there is a program for rating bearform tanking items with an interface similar to what you described above. (Getting character info from the armory, searching for upgrades, changing items slot by slot, changing gems/enchants, etc) I don't know, maybe you could talk to the creator of that program and try to save some work on the gui.
The Druid Wiki � Rawr

Aldriana drop Tornhoof a pm, he currently is already working on a simulator and gear optimizer using some "genetic algorithm". I have no idea what that is, but it's nerdy enough for me.

[Math models] Request for assistance

Just gonna squeeze this in since I didn't find much information about armors etc. Currently I just acquired 4 piece Tier 5 (chance on attack to make next move cost no energy). I was just wondering how much DPS am I getting from the proc, which I might add doesn't proc often (still need to check proc rate though). And how much better is 4 piece T5 from 2 piece T4, cause it seems im getting better rotations with the 2 piece T4. Thanks.

On the subject of balancing hit/ap/crit, I'm currently in a mix of T5 + BT/Hyjal non-set epics. I'm currently at 342 hit and 1857 ap unbuffed and I'm still a bit concerned hit is being weighted a bit too heavy.

Armory: The World of Warcraft Armory

"Better" is highly subjective. T5 4pc won't allow you to do 2s/5r the way T4 2pc does, but it is a very potent set bonus nonetheless, completely worth running a longer cycle like 5s/5r.

The key realization about 4/5 T5 is that, unlike T4 2/5, you're actually getting *more* energy to spend. 2/5 T4 lets you spend your energy in a more efficient fashion, but you still have the same amount of total energy. 4/5 T5 actually gives you more energy directly, and thus more damage - even if the cycle isn't quite as efficient.

In terms of how much it's actually worth: the Rogue Gear Spreadsheet includes an estimate for it. Typically the answer is "as soon as you can get it, you should use it, and you should keep using it until you have 2/5 T6"

Okay, so 2pcs T4 = energy efficiency / 4pcs T5 = more energy. Yeah I did notice some extra 25 energy procs when I do 5 point finishers. By the way, whats the most efficient rotation for 4pcs T5? I'm not sure which should I prioritize, Rupture uptime or SnD uptime? I try to keep SnD up always, but im just afraid of losing those rupture ticks.

If I get 2/5 T6, I guess im gonna use 2/5 T4 with it right?

It is basically always more important to keep 100% SnD uptime than anything else. If you worked hard enough coming up with a wierd gear combination you might be able to find one where that isn't true, but it's a good rule of thumb. SnD generates more damage per unit time than Rupture does.

Generally not, actually, for two reasons:

1) By the time you're at 2/5 T6, you generally have T5-equivalent gear in almost all slots; and for most people, dropping two pieces of T5 for T4 is a larger damage loss than what you gain from T4 2/5.

2) As your gear level increases into the T6 range, the value of T4 2/5 decreases.

Now, if you're in a situation where, for instance, you've picked up T6 gloves and helm but are still using T4 shoulders since Void Reaver hasn't cooperated, you could imagine using T4 legs for the set bonus until you got a shoulder upgrade. But this is pretty rare, on the whole; as a general rule of thumb, using 2/5 T4 until 4/5 T5 and then abandoning it forever is the way to go.

I'd also encourage you to work through the Rogue Gear Spreadsheet a bit; for whatever gear you're currently wearing, it will tell you the exact upgrade provided by each piece of gear and each set bonus, allowing you to work this out exactly for your particular case.

Quick question for you long time Sword players. Recently I switched over from daggers to swords and i am finding one problem. Every once in awhile, without AR, i will Build up 5 Combo points and still have a good amount of time of my rupture still up. Lets say between 4 and 6 seconds. So my question is what do you guys do? Do you keep SSIng until the rupture finishes or do you just auto attack then uses the rupture as soon as it drops? This is a problem you never really have as a dagger rogue.

I do one of two things. I either pop the rupture (if it will let me) early to help maintain my SnD cycle or sometimes, if the fight is very static (morogrim, winterchill), I just SS until the rupture is down and lose combo points essntially. \

It's usually a case-by-case decision, what I'm feeling in the fight and how secure I am in keeping my SnD up the whole time. I'd say "do what feels right to you right then". Cause sometimes popping that rupture early to maintain your SnD will seem right, and sometimes just waiting on it will seem right too.

*I know, not really an answer per se, but it's reality for me at least*

In case you accumulated enough CP and SnD/Rupture is still up, i would just auto-attack allowing the energy to replenish. By the time the energy reaches 90%+ a finisher should end and you start the next cycle with full bar being on the safe side in case Combat Potency will not proc too often.

And if you have consecutive lucky combat potenty cycles than:
1. Spam SS
2. play the lottery

PS - If you have 2/5 of T4 than 1CP for SnD is most of the time enough for sustained SnD and 5 combo Finisher.
I might just use 2 points for SnD in the cycle case you brought up.

I'd just remind people of the fundamental rules of rogue DPS:

1) Never let SnD go down
2) Never let your energy cap out.

So, if you can afford to autoattack for a few seconds without your energy capping out, you should do that. If, on the other hand, your energy is within a tick of capping out, you have your 5 CPs, and Rupture is still ticking (which, honestly, should be pretty rare), it is generally probably best to refresh the rupture (assuming you're doing a 2/5 T4-type cycle, i.e. 1s5r, which you pretty much have to be if this is coming up... if somehow you get into this situation doing 3s5r or 4s5r, the answer is different). The other reasonable alternative is to toss an extra SS, but my gut feeling is that that is inferior.

If you're regularly and persistently having this problem, what you probably want to do is switch to 2s5r and burn the extra SnD time.

It depends on how close you want to cut it, and how good your latency is. I mean, theoretically: if you hit the finisher the instant your energy ticks over 80, you have time to launch the first SS of the next cycle before the next energy tick, so you never clip any ticks; that is:

0.0: Energy ticks up to 90
0.1: Rupture for 25 energy; Relentless Strikes for 25 energy - still at 90 energy
1.1: Global cooldown back up
1.2: First SS of new cycle; 50 energy.
2.0: Next energy tick; energy at 70

Now, the issue here, of course, is that Combat Potency can screw this up; if you get a combat potency tick between 0.0 and 1.2, *that* gets clipped. So in practice, letting your energy tick above 80 while waiting for a finisher is a bad idea with combat potency (or if you have 4/5 T5, or whatever). But if you've been mashing and your energy is at 10 when you get your 5th combo point... you can still afford to wait 6 seconds, letting your energy tick up to 70, and then perform the above manuever, such that even if you get 2 combat potency procs in the 1.2 second window, you're still fine. And 6 seconds of autoattack will easily solve any rupture problem you may have.

Wait and autoattack. Your energy buffer takes a full 10 seconds to fill up from empty. You should easily be able to absorb 4-6 seconds "wait time" without any loss of energy. Hit the rupture at the right time, and you'll have energy free to spam SS and drain your pool down again.

If you're constantly finding yourself with a full energy pool and the need to wait for rupture, then you have a problem. Best option is to use an extra SS and waste the combo point - it's better than letting your energy tick beyond 100% and thus losing energy. If this is happening to you often, check out the DPS spreadsheet or gear spreadsheet (or both) and make sure you're using the optimal cycle.

Just to add to what Ald said (which I completely agree with).. If you are using the BT exalted trinket - this works out even better in your favor. You now can squeeze more SS's into your crit-proc time.

And yea.. whatever you do, the biggest things by far are: (as Ald mentioned a few posts above)
1) Never let SnD go down
2) Never let your energy cap out.

If you can afford to auto attack without breaking rule #1 or #2, do it and rupture as soon as the old one ends.
If you can't afford to auto attack (energy nearly capped).. check your SnD timer. If you are in danger of breaking rule #1, rupture if you can (if it won't let you.. less AP now than when you applied the first rupture) you'll likely have to evis . If you aren't in danger of breaking rule #1, I generally wait until ~80 energy and then pop in an extra (6th) SS. Basically treating it as an "excess energy burnoff" - similar to a warrior tank Heroic Striking to burn off extra rage. Perhaps doing the rupture thing early even if you aren't in danger of dropping SnD will be better - I'm not sure.. but it won't be a big difference either way. Just make sure rule #1 and #2 are followed.

Hopefully that was readable....

As an [Ashtongue Talisman of Lethality] owner, I absolutely love some energy regeneration time before renewing my 5pt SnD. This allows me to store energy until after the finisher so that I can get more sinister strikes in during the trinket buff. The rest of my cycle is essentially a race to get my 5pt rupture off and build up another 5 points for SnD with, again, enough time to spare for some regenerating ticks. This micro-management of energy use technique for Ashtongue users has been mentioned elsewhere in this thread (I think) as well as the Rogue Gear Spreadsheet thread.

I don't know if this was asked before or mentioned somewhere and I missed it (I didn't read through all pages sorry)

Does the comparison between Rupture and Evisc/Envenom assume Imp Evisc or Vile Poisons?

Also what would the numbers be like if crit rating was taken into consideration for the instant attacks? For instance if we calculate the average damage from 10 attacks assuming a specific crit %?
Would it then make sense to derive a threshold crit % which would make it more profitable on average to use eviscerate or envenom?

*EDIT: I wrote "imp poisons" instead of "vile poisons"

Bolded the appropriate section under Eviscerate, and re-quoted the Envenom section to emphasize the fact that at essentially any level of gear, it makes no sense at all to compare Envenom against Rupture. I did not factor in Vile Poisons in comparing Eviscerate against Envenom, because I don't think it's a worthwhile comparison to make. Note that for each point you spend in Vile Poisons, the damage of your Envenom increases, but so does the damage you lose from having to refresh your Deadly Poison stacks afterwards. I'm sure there's a point where Envenom's damage increase over Eviscerate scales to the point where it's worthwhile to use it, but as it is Eviscerate/Envenom are highly situational finishers. The build that would primarily be concerned about this question rarely seems to be run in raids anymore, too.