Wow, I totally missed that. Thanks.

I find it a bit weird that you guys are talkign about Don Alejandro's Money Belt and Belt of One-Hundred Deaths.

I just passed on Belt of One-Hundred Deaths because this spreadsheet comfirmed my own comparison with Belt of Deep Shadow. Belt of Deep Shadow is for me just as good as either belt but is dkp free and crafted pretyt much the second you step into SSC. It's also alot better then anything you get before entering SSC so everybody is bound to have this piece of craftwork. In fact there isn't a single belt that scores significantly better then Belt of Deep Shadow according to this spreadsheet.

Is there any reason you guys completely ignore this Belt of Deep Shadow in your comparisons of it's 2 alternatives?

Me and Aldriana were talking about our respective end-game "want" gear (Which are identical I guess in regards of most things.) With the exception of the belt.

Deep Shadow is really good though. But for my dream gear I'd want One-Hundred Deaths (Mainly because I mostly play as mutilate.)

On another note. I've so far put spending my arena points but this week I'll be reaching the cap and nothing else to buy.
So I'm torn between sword or fist. Post-patch Fist+OH Sword would be the best combination I guess?

Belt of Deep Shadows is a perfectly fine belt, and it probably is worth wearing for quite a while. But when defining the absolute best-in-slot item, even though Don Alejandro's and the Vashj belt don't beat it by a lot - they *do* beat it. So while it's not a huge upgrade and it's not worth spending a ton of DKP on or anything, if you're speaking of getting the best possible gear set you can, eventually you will want to replace Belt of Deep Shadows with one of the two discussed options.

I've yet to see or compute anything that makes me think first/sword is in any way superior to sword/sword. In fact, everything I've been able to piece together myself shows it as inferior, and I'm not aware of anything in the patch that would change that. So, myself, I'd be more inclined to sword/sword, unless I'm missing something important.

I remember some discussion over it in the other spreadsheet topic. The conclusion there was that fist/sword was very slightly better than sword/sword. For some reason I had that in my mind. Oh well. I'll decide next Tuesday I guess.

Well, according to this calculations in this sheet, assuming all other talents are equal, they're within a couple of DPS of each other. When you throw in that you have to take 5 talent points out of other talents - at least 4 of which can be providing DPS, usually via Vile Poison - knocks Fist/Sword behind, in my estimation.

Mmm, in my spreadsheet Dons belt is by far the worst belt at my current gear level dropping my dps by 4 over belt of deep shadows which in turn is only 0,1dps behind 100-deaths.

I also tried plugging in gear from MH and BT without actualy grabbing any tier 6 stuff (I won't soon see that as we're on Kael) and this still remains the same: bels of deep shadows is still better then dons belt although 100-belt overtakes by 1.5 dps now.
Only when I plug in the very best of the best endgear dons bels overtakes deep shadow while 100-belt stays the winner. (EDIT: meh, kinda forgot gems and a few pieces.. don wins in the end over 100 and deep shadow)

Then again... blizz just HAS to 'fix' Shadow-walker's Cord right? :o

On another subject.. this spreadsheet puts vile poisons in favour over imp poisons. In practice I find myself running around too often to justify speccing for vile. In an ideal situation (as a spreadsheet) I can see vile being better but in practice where often a deadly stack will wear of I still think imp poison is the better choice.
Anyone care to share their insights on this?

I've always gone for 2 points in each when i run 20/41 and 3 points in vile if im running 19/42

Well, here's the argument on belts as I see it:

Either you only care about raw DPS, or you care about the weighted total of stamina + avoidance + DPS.

In the first case, Belt of One Hundred Deaths does more DPS than Belt of Deep Shadows. In the second, the high stamina on Don Alejandro's puts it in the lead. So while Belt of Deep Shadows may have more DPS than Don Alejandro's, if you're worrying about DPS alone you shouldn't be using either. Fundementally: depending on your exact preferences you can justify using either Don Alejandro's or One Hundred Deaths; but regardless of your exact preferences, Deep Shadows will not be optimal by any reasonable defenition.

So, the sheet prefers vile poisons, simply because... it's modeling endless combat, so issues like the one you raise never enter the equation. In practice, on a more interrupted fight, you could make a case for imp poison being better. The general rule of thumb, though, is that if a fight is worth using Deadly Poison at all on, it's probably worth having Vile Poisons rather than Improved, as the faster stacking from Imp Poisons is only going to overtake the extra damage from Vile if the fight is less than 20 seconds or so long, at which point Deadly is sort of a questionable choice in the first place. Also, I would argue that there's very few 20 sec fights where your damage output on the level of poison selection matters. So, personally, I prefer Vile.

First, thank you so much for this site and the sheets. Sorry for the slightly off-topic post but after looking it seemed this was the best place for questions about the Gear Spreadsheet:

I got Warpspring Coil last night. The DPS spreadsheet tells me to replace Romulo's Poison Vial. The Gear Spreadsheet tells me to bank Bloodlust Brooch. Why the discrepancy? I'm 19/42/0.

I have a question regarding calculation of uptime for trinkes with cooldown:

Should that not be caluated as
uptime = Duration /( Cooldown + 1/(PPS *HPS) )
or aequivalent
uptime = (Duration *PPS)/(Cooldown*PPS +1/HPS)

HPS=total hits/sec.
PPS= procs per sec calculated based on hasted weapon speed
1/(PPS*HPS) is the time to next proc.

Do I miss something here? (...again )

The DPS sheet assumes a passive equivalent of Warpspring's proc. The Gear sheet models it as a proc.

Many thanks for this spreadsheet. It's been wonderful as my guild starts getting into SSC/TK, especially for evaluating marginal upgrades.

Also, a question: Everything I've read, including these forums and the Rogue Gear Spreadsheet, say that 308 +hit is required not to miss. I have 310 +hit, 2/2 Weapon Expertise and 5/5 Precision, yet I still miss ~6% of the time. (Actual numbers vary from 5.7% - 6.1%; these are provided by both Recap and my guild's WWS logs.) I figure I must be overlooking something obvious, but I can't work out what it is. Does anybody have any thoughts on what may be causing me to miss so much?

Cheers.

Click on the melee line of the wws report and see how much of that reported "miss" is parry and dodge. Recrap exhibits similar behavior.

I run with 229 hit 364 skill all the talents and only miss about 4% of the time.

Well, I hate to just say "because I'm right and they're wrong", so lets see if I can be a little more helpful...

On any particular item, the two sheets will often disagree on the value of certain items. The reason for this is that the two sheets have very different models for rogue DPS, and thus have different levels of accuracy for different effects. On the whole, I believe this sheet to be more accurate than the other (as if I didn't, I wouldn't have bothered to write it). Ultimately, you need to decide which sheet you trust more.

In the specific case of RPV: I believe that, on the whole, this sheet tends to value +hit more than the other, due to the fact that I model all procs as procs rather than static bonuses, which thus includes the subtlety "you cannot proc unless you hit", which understandably drives up the value of hit. This is probably at least some of the difference.

However, it is also possible that the two sheets have different values for the proc rate on RPV entered; in this case, I can't actually claim with certainty that I have the correct values, as while I have done some rough testing myself, I am not aware of any firm conclusion being reached (or at least, there's no ultimate conclusion found in the [Rogue] Mechanics Testing Thread). As such, while I believe my estimate of it's proc rate to be in the correct ballpark, it is entirely possible that it is misestimating the amount to some extent, which may explain some part of the discrepancy in the sheets.

So, I'm not exactly sure what you're objecting to here, so I'm going to address both things:

1) With regards to hasted weapon speed consideration: this is currently not modeled in this sheet, as the last revision occurred before the new findings regarding weapon speed and haste. This update will be in the next version.

2) With regards to the extra term in the calculations: that is done on the line before. By "procs per second" I mean "the actual probability that a proc will occur during any given second", which is computed as hits*procrate, in the previous line.

Recap and WWS's overall hit number counts dodges, parries, and blocks as misses. If you're attacking from behind, as usual, this works out to about 5.5% of attacks missed to these effects. As such, your miss rate, even without any actual "misses" with show up as 5% or more in many cases. However, if you look at the detailed breakdown with either program, it will correctly break it down into hits, dodges, etc.

So: you're probably not actually missing that much. You're just getting dodged that much, and, alas, there's no good way to fix that.

I haven't seen sufficient evidence to totally convince me that instant attacks don't suffer from the reduced proc rate. And, really, while it does make proc effects depend less on weapon speed, it doesn't actually allow the calculations to be simplified at all as proc uptime still does depend on haste via the haste -> more CP procs -> most instant attacks -> more procs feedback loop

But, yeah, implementing the new findings won't be too hard - I've already started work on it, actually. I just don't want to start filling in the details until they finalize the behavior in the patch.

I'm not sure if you're aware of Excel's stupid powerful iterative calculation feature, but I started using it in my own spreadsheet and found it allowed me to accurately model the haste feedback loop. Basically you can set up a circular reference in some calculation in your spreadsheet, and Excel will give you an error -- then go to the settings dialog to the "Calculations" tab and turn on Iteration and BAM, it recalculates your circular reference down to within a threshold you can specify.

The only problem I've found is that if you get particularly fancy with it, it can break (at which point you'd need to cut off the circular link at some point so that Excel can figure out a new starting point to recalculate). But I'd think that, based on the way your sheet is set up, it would work without any complaints.

Regarding yellow attacks and haste, did you see these two posts?

http://elitistjerks.com/479974-post120.html
http://elitistjerks.com/427220-post28.html

(edit) One snag, I don't know if OpenOffice has a similar feature. Anyone know if it does?

My sense is that macros in Excel and macros in OO aren't compatable, which is why I'm not using them. That said, the manual-loop method that I'm using works just fine; all I'm saying is that the changes, while they affect the rate of convergeance of the loop, they do not eliminate the need for it.

Regarding yellow attacks and haste: Yes, I saw those posts. Here's my counterargument:

1) Fiery does not seem to reduce in proc rate with haste as the expansion procs do; as such, the fact that it's proc rate on instants is not reduced by haste doesn't mean a whole heck of a lot.

2) The proc rate of RPV was barely reduced on white attacks by haste; since some procs seems to be reduced and others not, there's no particuar reason to assume that it has to be. Thus, I would want to see a larger data set showing conclusively that RPV proc rate is reduced by haste before accepting that data.

3) Hence, the only one of those 3 data sets that I put much faith in at all is the third, which, while it is a reasonable test of a proc who's proc rate has been shown to drop via haste, is still flawed; the data set is far too small to show that there's conclusively a difference.

So: while I do agree it's more likely, based on that data, that some PPM effects are reduced by haste, I don't think it has been proved beyond reasonable doubt.

As an additional note: given that some procs exhibit proc reduction rate with haste and some do not, it would probably be worth testing more different procs to classify which do and which do not, which has not been done to the best of my knowledge.

Yes I am aware of this.
After you mentioned it, I can see it now. I dont know why I did not caught it earlier.
I asked because am always suspicious about seemingly random constants in formulas

Yes it has.
You find it under Tools->Options. This pops up a frame where you go to
OpenOffice.org.Calc -> Calculate.
In the top left corner of the pane you find it.
Beware, OO limits the number of iterations to 1000 (Excel allows 32767)

wow, I'm pretty obsessive about checking this thread every day, but I just read walloftextwalloftextwalloftechnicaltext.

/nod

/walk away