• Vulajin

Current version: 0.4.7 LK (02/15/09)
BC legacy version: 0.4.1 BC
Download links: Attached at the bottom of this post.

(NEW 02/18/09) I was able to achieve X DPS on the spreadsheet! Nobody cares unless your new high score uses some new type of cycle others haven't yet considered, or uses a gear selection that hasn't yet been discussed in the T7.25 Rogue Gear Discussion thread. If you're simply posting to show how you "won" the spreadsheet, you'll get an infraction.

(NEW 02/16/09) How the hell do I use this thing? A poster over on the Wowhead forums put together a helpful thread with instructions for using the spreadsheet. You can find it here: (link)

What is it? The Roguecraft Spreadsheet is something I have been working on and using for my personal theorycrafting efforts since last October when I wrote Roguecraft 101. Remember those posts I made in the Hemo thread back around the time of patch 2.3? All of those conclusions were based on a previous version of the Roguecraft Spreadsheet. The sheet has undergone several complete overhauls since then, improving in accuracy and usefulness each time. Its calculations represent our best current knowledge of Roguecraft, and they are designed in such fashion that I can easily extend them (for the most part) to incorporate new features.

What does it model? The Roguecraft Spreadsheet models all current DPS-affecting talents. It models all common types of cycles used by the common builds, and it models Seal Fate's effect on those cycles. It models all current gear and trinkets, including our best current knowledge of all procs. It models all raid buffs and debuffs and all cooldown abilities. Any of the above that have been omitted were unintentional and should be reported in this thread.

So really, what's special about it? This is the key question. Why, really, did I feel the need to release a spreadsheet I have been using solely for personal use for over 8 months? The answer is a bit complicated, but it basically boils down to this: as the author of Rogue: PvE DPS (and previously Roguecraft 101), it is very important for me to be able to quickly and accurately assess the value of a newly proposed ability or talent or change to an existing ability or talent. Basically, I wanted the accuracy of the Gear Sheet, but I wanted much greater extensibility and, in addition, I wanted to build from the ground up to be inclusive of Seal Fate specs with the same degree of accuracy.

The sheet also contains two particular features I'm very proud of, neither of which currently exists in the other spreadsheets:

- Energy pooling. You can specify how much energy you wish to pool before performing each finisher in your cycle. Depending on how much energy you pool before a finisher, the number of combo point builders you can use while under the effects of [Ashtongue Talisman of Lethality] or Find Weakness will vary, and in addition if you do not pool enough energy before your Slice, you may not be able to use your next finisher during Exploit Weakness or Find Weakness. The Roguecraft Spreadsheet models both of these effects.

- Cooldown timing. You can specify exactly when to use your cooldowns, including Drums and Bloodlust. This feature can be used to model the effect of stacking cooldowns rather than simply using them haphazardly. You can also specify a particular fight duration (maximum of 600 seconds, or 10 minutes) to reflect a specific fight (for example, 360 seconds for Brutallus). Of course, if it's not worth the effort, you can just set the fight duration to 0 seconds and it assumes time-averaged effects for all of your cooldowns, as both existing spreadsheets do.

How do I use it? Basically, the Roguecraft Spreadsheet was never really designed with any users other than me in mind. So, basically the UI is generally accessible but not especially user-friendly. Frankly, I don't think I will change that too much, because I'm far more concerned with the actual accuracy and quality of calculations than I am with the UI. I have no intention to ever include features like Armory Import, or saving talent builds, or things of that ilk. There are also currently no readouts for EP values or anything of that sort. It's very minimal: there is a talent selection sheet with some general settings, a gear selection sheet that should appear familiar to most users of other spreadsheets, a buff selection sheet, and a cooldown settings sheet. All of the calculation sheets are hidden by default; you can find the iterative calculations on "DPS Calc 1," "DPS Calc 2," and "DPS Calc 3," and the corresponding numbered sheets for "Cycles" and "Cooldowns" as well. The final calculations are on "DPS" and "Cycles." All of the gear data is contained in the respectively-named sheets.

Be careful with the sheet, as there is not really any error-checking and it is fragile. Set up your build on the "Talents" sheet, then select a combo point builder (even if you take the Mutilate or Hemorrhage talent, you must specify Mutilate or Hemorrhage as your CP builder; conversely, you can specify one of those as your CP builder even if you don't have the talent for it), a cycle type, and the number of CP to use on each of your finishers. If you wish to pool energy prior to a particular finisher, you can specify that in the column to the right of the cell indicating the number of CP for that finisher. Sorry, but the sheet is not equipped to recommend cycles or gear for you. If you're unsure what cycle to use, you should read Rogue: PvE DPS for some ideas, or just play around until you find something that achieves high DPS. If you wish to take advantage of the more accurate cooldown-stacking modeling without having to specify every single use of your various cooldowns, then set a fight duration longer than 0, and simply change all the yellow-highlighted settings on the "Cooldowns" sheet to FALSE.

What if the sheet disagrees with the two existing spreadsheets? In the case of Aldriana's Gear Spreadsheet, several discoveries have been made in rogue mechanics since Aldriana discontinued work on the sheet. In either case, there are several ways in which this sheet's models differ from those of the existing sheets which may result in differences in its output. As a biased observer, I regard my own sheet as the most accurate currently available. However, if you would like to discuss any differences you find while using the sheet, please utilize this thread for that purpose.

If you have any other comments or questions about using the Roguecraft Spreadsheet, please post them in this thread and I will do my best to answer them.

(This spreadsheet is dedicated to Xiah of <Shattered Oath> on Emerald Dream (US), for always being awake while I was fiddling with this damn thing and making idle conversation with me.)

Attached Files

	Roguecraft BC 0.4.1.xls (4.58 MB, 19220 views)
	Roguecraft LK 0.4.7.xls (3.35 MB, 77706 views)

 

Capek

Thanks for this. Always going to be interesting to see the backend to all of your work.

Immediately of note is that Fleet Footed has 3 possible points in your list but actually only has 2.
Currently that doesn't matter as Fleet Footed is having no effect on DPS. Neither is the speed buff of Cat's Swiftness (tested by setting to 'none' and manually adding 6agi).
Is there any chance the DPS increase garnered from increased movement could be implemented? Perhaps associated with the 'combat duration' tab whereby we can specify an estimated amount of time spent travelling (Gruul springs to mind as the simplest example).

 

Wynna

So far what I've noticed is that your DPS doesn't change at all no matter how many Bloodlusts you specify, from either 0 to say, 4.

Also thanks for modeling AtoL so well, it is apparently quite underrated in the other sheets.

 

• Vulajin

Regarding the Bloodlust issue, not losing DPS when you set it to 0 is definitely a bug, but if you set it to 4, and if you have "specify Bloodlust timing" turned on in the "Cooldowns" sheet, you would still need to specify a time to actually use that 4th Lust before it would have any effect. I'll look more into this.

The thing here is that this gets into modeling the exact number and nature of interruptions during a fight. This is not really the province of a spreadsheet; as it is, the formulas to implement the cooldown stacking model are annoyingly complex, to the point that the "Cooldowns 1," "Cooldowns 2," and "Cooldowns 3" sheets occupy 1.6 mb of disk space by themselves. I expect that Aldriana is looking at implementing some sort of interruption model in RogueCalc, which would be a much more sensible thing to do. I would suggest that for now you use the formula given in Rogue: PvE DPS for comparing Cat's Swiftness to other boot enchants.

 

• Aldriana

The challenge of movement-based effects is that some abilities/effects - particularly trinket procs - are very highly subject to the exact interruption pattern of the fight. While 30 sec in/10 sec out and 45 sec in/15 sec out will be more or less the same in terms of the value of hit, crit, ap, and so on, they will *radically* change the value of some trinkets. Moreover, there's complications wherein some fights have interruptions that Cat's Swiftness helps with (Gruul, Archimonde), while others do not (Naj'entus comes to mind). So interrupted modeling requires very specific information about the timing and nature of interruptions... never mind the fact that we have no real idea what the optimal cycles for interrupted combat look like in the first place.

Also, part of the benefit of effects like Fleet Footed is not even in their direct damage contribution. On a fight like Felmyst the extra run speed can help get you out of the way of his various attacks more quickly, which has no implications to DPS but large implications to suvivability; hence, the exact DPS benefit would not be a notably helpful quantity in terms of assessing value.

Hence, while ultimately it would be helpful to have that sort of modeling in place, I don't think we understand the problems involved without going all the way too writing custom code for each strategy of each fight - which I for one have no interest in doing.

I think what it comes down to in the end is: while spreadsheets and the like are a valuable tool and provide a good baseline for your analysis, at some level you still need to know what you're doing, as there are some subtleties that just cannot reasonably be modeled. And the value of runspeed, for the moment, falls in this category.

 

Vodrin

I love the AtoL modelling although it seems to show inflated levels of dps, 2600~ for badge fist + savagery all buffed up (1 heroism) for brutallus but netherless has been useful on the energy pooling/AtoL side. Thanks :>

 

• Aldriana

One shouldn't be observing large differences in Ashtongue quality due to energy pooling - it'll be a little better, certainly, but, fundamentally speaking: the Gear Sheet tends to estimate ~70% AToL uptime for SS builds; the damage difference between that and 100% uptime is only on the order of 40 DPS. Since in practice uptime that high is not achievable via energy queuing, the observed difference should be less than that - I believe I one estimated it to be closer to 10 DPS. So if people are observing differences larger than that, it's very likely a bug.

 

Vodrin

Actually the energy pooling shouldn't affect uptime much at all, only the ss/hemo that fall into the uptime?

Thats indeed how the sheet changes too when modifying pooled energy and it is only a 10dps change. This sheet is modelling it as 76% uptime for my gear though. I don't have dst or wsc though so its important to me :p.

 

• Aldriana

Well, right, that's exactly the point. The largest possible advantage is 40 DPS; but since you're really only increasing uptime on yellow attacks (which make up perhaps 30% of your damage) the observed benefit should be closer to 30% of 40 DPS, which is 12 DPS; and even that assumes that every single yellow attack lands during the buff uptime, which, in my experience, is not the case - you get it up to 90%, perhaps, but every once in a while things don't work out and you miss one; and if it's only 90% rather than 10%, your benefit is down to 8 DPS. So while, I'd expect Ashtongue Talisman to be a bit higher, but I wouldn't expect to see any real night-and-day differences with it.

 

• Feist-Mok

Doesn't work in Office 2008 under Mac OS X, but neither does the most recent version of Aldriannas sheet. I try to change one piece of gear and it locks up. So, I'll chalk it up to an office issue and give OpenOffice a whirl when I get home.

 

• Vulajin

In actuality, at least according to my spreadsheet's results, the difference is even less than that. In other words, single-digit DPS gains from pooling in conjunction with AToL. However, I also found that the base value of AToL tended to be a bit higher than what was previously modeled, regardless of energy pooling.

 

Wickedchild

Tunic of the Dark Hour is missing its gem socket.
Trying to add it through the Chest sheet reveals that the stamina cell for the Tunic is set at 51+3*M10, M10 being the Socket1 cell for the Tunic, which is empty, so adding Red in there gives an error obviously.
I simply removed the "+3*M10" to remedy this.

 

• Vulajin

Yeah, that's a relic from an old version of the sheet. Your fix would be the correct way to temporarily patch it. I'm not going to be too quick about releasing 0.1.2 unless bugs of significant substance pop up. My chief priority right now is the actual accuracy of calculations.

 

Khana

Why is no spreadsheet modelling shadowstep?
It can be considered a timed buff with a 1 minutes cooldown to insert into any standard rotation, possibly on ruptures. 1 Rupture cycle every minute is buffed. I think there is still a lot to be said on this, I was able to see some kind of 600 ruptures tick with Mangle up to the target.

Hemo is quite the same as well, it can be reduced as a buff. Any Hemo specced rogue spams hemo, so we can considered it a permanent +DMG buff on any kind of ability it can be applied to and instead modelling a "debuff" on the first strike, when it is not yet in place.

 

drumbum

Napkin math suggests that Shadowstep is always a DPS decrease, except possibly when used with Garrote.

Consider that Shadowstep costs 10 energy. Thus, you can increase the damage of your rupture by 20% for an energy cost of 10. Is it worth 10 energy to deal 20% of Rupture's damage? A similar question would be: would you spend 50 energy to deal 100% of Rupture's damage? I doubt it.

 

Khana

Normaly a rogue who has Shadowstep also has Serrated Blades, so the two %buffs stack.
Which means that upon activating ShS you'll get [Rupture DMG] + 30% + 20%.
I don't know where Mangle buff stacks in, I mean, at which point.
Considering also that to go down to Shadowstep you have to take Sinister Calling and Deadliness, the AP buff you get will boost those "basic" Ruptures. So it's not only about 10 energy for a +20% damage, it's a 10 energy for a (yet to calculate) improve of absolute rupture damage. Damage that comes at a 50% threat. This threat dump works better than feint, which is widely known and proved to be kind of "broken" and useless.

Clearly, that's for the sake of calculations, as a full subtlety build going down to shadowstep lacks the energy mitigators needed for sustained fights. But it's still an option for raids up to Gruul.

 

Andeh

I noticed some weirdness with Bloodlust timing. If timing is left unspecified for 2 Bloodlusts, it comes out *significantly* higher than if you specify the timing for the end of a 6 minute fight. Significantly as in on the order of 200 DPS. In this case, I set the first one to 270 seconds, and second to 270+55. I made sure to adjust 2 minute cooldowns such that third Blade Flurry would overlap the first Bloodlust (270).

1) Any explanation of why this is happening?
2) Is the difference between "time averaged" and "time specific actual modeling" that big?

 

• Aldriana

So, I don't think Shadowstep is totally useless, but lets take a minute to think about the situation.

1) Using SnD on a CPG (Backstab, Hemo, etc.). Well, this increases the damage of that move by 20%, at the cost of 10 energy; hence, it only improves the energy efficiency of the move if the move in question costs at least 50 energy to start with; hence, it never makes sense to Shadowstep a Hemo.

2) Shadowstepping a Backstab will be marginally worthwhile from a straight damage-per-energy analysis, but only barely - assuming your backstab hits for an average of, oh, say, 2100 damage, your damage per energy goes from 35 to 36. Thus, the net damage gain from the Shadowstep is about 70 damage. But Shadowstepping in this way also costs us 1/6 of a combo point over the long run, and a combo point spent on rupture is worth about 400 damage. Hence, the loss of the combo point *costs* us about 70 damage. Thus, shadowstepping a backstab is more or less zero sum.

3) Shadowstepping a Rupture. Well, this is very likely going to be worth it - lets run the numbers. Assuming a 2500 damage Rupture, we get an extra 500 damage for our 10 energy; that 10 energy, spent on a CPG, would give returns on the order of 400 damage counting both the direct damage and the resulting combo point. As such, we gain maybe 100 damage by Shadowstepping a Rupture, which, at one shadowstep per 30 seconds, means we gain perhaps 3 DPS.

Thus, the reason why no one bothers to model shadowstep is because 1) it's sort of a pain to model, and 2) the damage contribution is just this side of totally irrelevant anyway.

As an additional note: feint is only useless if your tank is good. With a low-threat tank (such as one I had in a prior guild who was only pulling about 500 TPS in SSC), feint is actually relatively useful. Similarly on aggro-limited fights, such as the early parts of Gurtogg. Which is not to say it doesn't need a buff - it most definitely does. But to declare it to be unilaterally useless is a mistake, in my opinion.

 

Khana

well the issue is that as long as we don't calculate, we don't know and I have the slight impression it is a bit more than 3 DPS, but it's a sensitive one, so don't take it into consideration.
Also, it's true it goes against a CPG, but considering that CPG would probably be Hemo (Shadowstep-raiding with daggers makes you only a great trash mobs killer, you desappear from any dmg list for fight about 5 minutes length - tested) we should value the damage loss of a single hemo each 30 seconds. Guess it is anyway less than 3 DPS.

Point 1) is sure proven, and I guess it's the only reason why no one model it.
Point 2) is in the field of speculations. Consider a Hemo shadowstepper with swords specc, with a 0/20/41 imp SnD and DW specc... a haste-rating set.. 2 T4 bonus set for 1s/5r rotation... standard rupture ticks for 500, buffed by ShS and then again buffed after 35% HPs by Dirty deeds.
You can say this is kind of strict to some specific situations, but well... it's a model that still lacks.
I remember Dr Damage telling me I had a 2400-something rupture damage on mid-KZ gear level, specced shadowstep.

Funny side note is that the only buffing talent is on the subtlety tree that no one uses in PvE

Well anyway *shrug* the point is only that I am curious about the reason why Blizzard put talents that directly buffs ruptures or can be used as rupture buffs in a tree that is mainly considered PvP related, not telling about the threat reduciton thing.
So there -should- be a way to properly use ShS in PvE...

 

Nock

So I was fiddling with this sheet, just seeing what it had to offer, and playing with different cycles.

The gear I have input is what's listed on my armory 95% of the time, so you can get that there. Buffs were standard, Shaman, Ret Pally, DPS warrior, 1 heroism, 3 drums. Standard debuffs.

The weird thing that I encountered was playing with your Xs/Ys/Zr setting. Using 3s/5r with the sheet, it shows me with 2687.74 dps. I had it set to 5s/3s/5r and it showed me with 2710.03 dps. Dropping it to 5s/1s/5r upped the number to 2719.17 dps.

Messed around with other cycles, and get some interesting numbers for Xs/Ys/Ze cycles, but nothing higher than my standard DPS cycle like that was.

 

• Aldriana

1) Don't split quotes. It annoys the moderators.

2) I suspect what Blizzard was thinking when they put in the threat reduction is that you would use this when first entering a fight (as it originally required you to be stealthed) - and if your opener was at risk of pulling aggro, there'd be no advantage over just running in as you'd have to wait to start attacking anyway. So I suspect the aggro reduction is there so you can shadowstep in right away and get started faster than your pals running in. Which is not a terrible idea, it's just such a small benefit that it's irrelevant in any fight that goes more than about 30 seconds.

You're also making the assumption that Blizzard actually knows what they're doing in terms of balancing some of these things, which I think is a bit of a leap of faith.

3) Okay, lets do some numbers on x/x/41 hemo builds. I did Backstab the first time through because that seems to be the way deep subtlety is moving next expansion. But, sure, lets do Hemo. Assuming you're using a 100 DPS 2.6 speed weapon, have ~3500 AP buffed, and say a 35% crit rate (reasonable for a T6 rogue), your Hemo hits for around 1050 noncrit and 2100 crit, for an average of 1450 or so. Assuming you have, say, 600 ArPen on gear against a sundered, faerie fired, and CoRed boss, mitigation reduces that by 20% or so, to, say, 1150 or so. But, it also provides a debuff that is worth about another 350 damage, so we're talking about 1500 damage or so all told, or about 43 damage per energy. Hence, the 10 energy spent on Shadowstep costs us 430 damage outright from the Hemo's damage, plus a bit more for the combo points.

Meanwhile, with our 3500 AP, a Rupture hits for base 1000 + 3500 * .24 = 1840 damage, which, with Mangle and Serrated Blades, becomes almost 2900. A 20% increase on this is 580 damage; thus, even neglecting the loss of 28% of a combo point due to using Hemo, our damage benefit is only 580-430 = 150, which works out to 5 DPS. And I'm willing to bet that the loss of a combo point (which is hard to quantize better than the earlier estimates) eats a way a fair bit of that.

Regardless: we're talking at most 5 DPS or so. I stand by my original assessment.

 

• Vulajin

Thanks, I'll investigate the s/s/r and s/s/e cycles to see what's up.

(edit) Found several major errors in the s/s/r and s/s/e cycles, as I expected. I knew I had forgotten to test something. Problems will be fixed and 0.1.2 will be released in the near future with these fixes. (edit 2) Same issues applied to the s/r/e cycles and have also been fixed.

You stumbled upon a nice little bug wherein the cooldown model automatically assumes 5 Bloodlusts (used sequentially at the start of the fight) if you set duration > 0 and don't "specify Bloodlust timing." It'll be fixed in 0.1.2.

The reason why Shadowstep and the Hemo damage buff are not modeled is because I forgot to before getting this out the door. Neither is of significant interest to most of the rogues for whom this spreadsheet is intended, so I won't be rushing out the next version with these things implemented, but they'll certainly be in 0.1.2 whenever that's ready.

 

• Vulajin

Version 0.1.2 is available via the link in the first post. Changes:

- Fleet Footed (Assassination) changed to 2 points maximum.
- Initiative (Subtlety) changed to 3 points maximum.
- Many fixes for three-finisher cycles (s/s/r, s/s/e, s/r/e).
- Setting duration > 0 and not "specify Bloodlust timing" no longer automatically assumes 5 Bloodlusts.
- Automatic Bloodlust timing now assumes one Bloodlust every 120 seconds, instead of all in a row at the start.
- Unhid the "DPS" and "Cycles" sheets by default.
- Streamlined several DPS calculations.
- Added Scroll of Agility V and Scroll of Strength V as consumable buffs.

(edit) Tunic of the Dark Hour still doesn't have its socket or bonus because I'm functionally retarded.

 

Wickedchild

Hardly worth a mention again, but I figure its better it doesnt slip by you, the Tunic still seems to be missing the gem and bonus. :P

 

• Vulajin

Err. I could've sworn I fixed it last night shortly after I read about it. I swear to god, I'll really fix it in 0.1.3. Until then, the fix is still detailed above.