I didn't update my Spreadsheet for awhile but getting some weird results when i did.
I'm combat swords spec

In my 2_3_2_5 spreadsheet Hourglass of the Unraveller was a -4.55 DPS downgrade from Blessings Deck : Crusade

In 2_4_0_5a spreadsheet it's showing a +5.20 DPS upgrade over Blessings Deck : Crusade

Was something incorrectly modeled in my earlier spreadsheet version or is there some sort of bug throwing off the new spreadsheet?

This 10 DPS swing is with identical gear, gems, spec etc except for the trinket swap itself.

Thanks for some clarification on this

My cycle changed as well from a 3S/5R to a 4S/5R snd cut

Not sure why the change here as well with identical gear??

Thanks

DMM made some changes to AR, which might be the reason, and I made som changes to Seal Fate modelling and Haste effects on cycles; either could be responsible for this difference.

Same happened for me, but 4s is more accurate i presume (I pretty often ran into trouble with 3s, even when starting slice with 70-80 buffer energy).

Just to report what I think is an issue with the OpenOffice version of the spreadsheet.
I use both Excel and OpenOffice spreadsheets, as I have Excel in my laptop at work and OpenOffice on my computer at home.

When the 2.4.0.5.a got out, I plugged my equipment and my buffs and got a value of 1624 dps. That's with Microsoft Excel.
Now, if I open the same spreadsheet with OpenOffice, I have a boot that takes me to 1785 dps ! So, I thought that the compatibility between the two applications was the cause of the boost and I downloaded the OpenOffice one (no macro buttons) and plugged in my equipement and my buffs, ending up with the same result : 1785 dps.
What's with the OpenOffice spreadsheet ?

EDIT : sorry, I found why by looking a few pages back in this thread (post 548). It's working ok now.

Going back through this thread as I now have time...

Cycles...
I recently removed Adrenaline Rush from the energy cycles and assigned it's damage separately. Previously it was adding the energy to your cycles and the cycles were calculated accordingly. This works with Haste potions as 40 energy can usually just get added to your cycle. 150 energy isn't able to be absorbed that way though because your energy would cap out if you maintain the same cycle. The result for Combat Swords builds looks to be cycles 1/2 a step higher, so 4s/5r cut from 3s/5r. These should be truer cycles. The previously suggested cycles (for builds with Adrenaline Rush) were probably a touch energy starved based upon the 150 energy that couldn't really be properly absorbed into your regular cycle. I may readdress AR more in the future (as most people probably add a full 5e to their cycle and "borrow" extra energy), but this should be much closer (and probably close enough that the prior mentioned tweak probably won't change the suggested cycle).

Hourglass vs. Crusade
Well, every version we learn more about the game mechanics or find ways to improve the modeling. My gut says that for most raiding rogues, Hourglass will be slightly better than Crusade but they are rather close. We'll see after the next version comes out as there is a little more find-tuning.

Hemo
As mentioned many times throughout the thread, the spreadsheet defaults to including an estimate for the hemo contribution to raid DPS. If you look at the FAQ (first page) it discusses this more in depth.

Next Version
I'm hoping to get this out today, but this time I'm going to test it a little more thoroughly. It will include the recently learned Slayer 4-pc changes, more tweaking to Extra Attacks, the correct crit damage formula is implemented (in fact, part of the sheet used Aldriana's crit formula, but I found this wasn't added to the cycle sheets yet). And the one thing I want to implement today is the Windfury buff AP "carryover".

Blade Flurry
That is not a bug. I moved Blade Flurry and merged it with permanent haste (as that's how it's used in the spreadsheet anyway). Essentially, that cell has been obsoleted. In previous versions that was SnD x Blade Flurry in that cell.

Unmodeled items
I actually expected people would type the actual name of the item where it now reads (Unmodeled). I will change that to Unmodeled MH Mace, etc. But as an FYI, you should be able to type the item name in the cell. It should still show up at the bottom of the list (not in alphabetical order).

Minor bug here: in the Ring Sheet, row 32 is repeated, so one of the rings shows up twice in the dropdown.

Good catch. The 2nd "unmodeled" ring is currently off the list and unaccessible.

I'm seeing an 80 DPS increase going from 05a to 06, of which 73 DPS is white DPS. This seems quite hefty, is this the result of the sword specs proccing off specials or did you change something else as well?

// Edit
Just noticed while searching for updates, for the first time DST has been pushed from being #1 trinket. For me, Shard of Contempt is now a slightly better upgrade than DST (yeah, I still don't have it...).

I was just wondering, is RogueDPS_2_4_0_6NA the latest version of this spreadsheet ?

_NA is the Mac-Version without armory import.
For Windows-MS-Office-User 2_4_0_6 is the latest version.

Here is a followup to my last post on a new method of Mutilate modeling. I've been in conversation w/ Todemax via PM's and he has explained to me more or less how the current Seal Fate modeling works in the current non-Mutilate cycles. (This is the new Seal Fate modeling, after it has been fixed to actually make sense.) What's there right now for the non-Mutilate cycles, as implemented by Todemax and DMM, is more or less what I was thinking of for updating the Mutilate cycles.

Basically, the current implementation is a X+/Y+ cycle, where the damage output is weighted by if you get additional combo points or not from your Seal Fate talent. The reason it hasn't been implemented for Mutilate is the difficulty of determining the percentage of the time that you would get the extra CP. I'm hoping to help fix that issue.

As a start, here is an updated set of state/path diagrams for Mutilate: Savefile.com

EDIT: Updated the URL to a new file with the error that Drumbum caught corrected.

EDIT #2: In the above file "Relentless Strikes" should read "Ruthlessness"... (I always get those two mixed up. However, it's not worth it to upload another file... the formulas don't change.)

(By the way, anyone know how to attach a file directly to your post? Do you need special privileges to do that? I would assume so.)

The last page of that set shows a table of the probabilities of different outcomes based on the parameters specified. The table separates out probabilities for numbers of combo points and numbers of instant attacks used. In practice, a spreadsheet would likely implement this a little differently. I'm going to try to put together a spreadsheet soon which demonstrates the concept I'm getting at.

As always, feedback welcomed.

Thanks for putting the state diagrams together Left. It makes it a lot easier to follow what is going on.

I did some error checking on the probability analysis tables and found an error in the formula for 2 instants in a 5 cycle. The red portion is the term you left off:

Everything else checked out. I look forward to seeing this implemented into the spreadsheet.

Thanks; nice catch. I'll correct my previous post with a fixed version.

That seems pretty much to be sword spec proccing off specials. It actually puts it back about where it was prior to 05 which indicates to me that the spreadsheet (in its own special way) was already modeling it that way and explains the relatively hefty DPS loss for Combat builds in 05. The change was necessary to more precisely implement Windfury primarily.

I haven't gotten to more precise modeling of DST yet as the Next Stat issue was causing me fits. Hence the remodeling of the white damage section. With the Next Stat section temporarily bypassed, I can think about that soon.

In the buffed miss percentage, is it 0.00% or is there a fraction still to be had? It might somehow be possible that due to rounding you are a sliver away from the actual hit cap. I tried a couple tests and was not able to replicate that issue, at least with the default combat swords build.

EDIT: Sorry about the wall of text. I'm trying to split it up a bit...

Ok, so, I have something now that can theoretically be integrated into the sheet as it currently works. Unfortunately, I can't figure out how to do the integration.

Proof-of-concept
Here is what I have developed: Mutilate Cycle Distributions

This is a proof-of-concept Excel file which shows how Seal Fate, Ruthlessness, the Mutilate attack, and 4pT4 set bonus interact to produce probabilities of different outcomes when using the Mutilate attack. The cells in the upper right just duplicate cells (and names, for the most part) from the actual DPS spreadsheet. They are basically placeholders for the corresponding cells in the DPS spreadsheet. Note cell F2, which calculates the instant crit chance.

The tan cells are tables which show the possible outcomes of a series of Mutilate attacks and the likelihood of reaching each outcome. The formulas for these tables are pulled directly from the state/path diagrams I drew up earlier (see my earlier post). At the top, I have cells with names "prob_r, prob_A, prob_b" and their corresponding opposite probabilities "prob_notr, prob_notA, prob_notB"; again, these correspond to the variable names I was using in the state/path diagrams.

Each of the other blocks of tan cells represents a cycle style: 1+, 2+, 3+, 4+ and 5. Basically, if you are aiming to get at least the number of combo points specified on the left side, then the other cells determine the amount of combo points you actually will get, on average. Of particular interest is the number of Mutilate attacks, on average, you will have to make to reach your target. This average is included at the bottom of each tan block, along with the average number of combo points generated by the cycle.

Next, the yellow boxes off to the right are designed to correspond to the existing yellow boxes in the DPS spreadsheet which are used for Seal Fate probability. The table in my spreadsheet entitled "Combo Points Reached" corresponds to the cell block in the DPS spreadsheet which begins at N9. The table in my spreadsheet entitled "Mutilates Needed" corresponds to the cell block in the DPS spreadsheet which begins at K14.

Implementation (issues)
Ok, so we've determined a way to calculate the probability of different outcomes. Now what? Well, this is where I'm stuck. I tried to insert the tables I created into the DPS spreadsheet and modify an existing cycle. Here is what I came up with: MODIFIED 2.4.0.6 Rogue DPS Spreadsheet.

After finessing the cell names and checking some stuff, it appears that the tables work ok using the source values (talents, crit %, etc) provided by the sheet. However, I couldn't modify the existing cycle easily... at all.

In the modified sheet linked above, my tables were inserted way over to the side, starting around cell X26. Once I had them working, I copied an existing, working DPS cycle (Row 67) so that I could try to modify it. The modified cycle is located in Row 142.

At this point, I still had a combat swords build, so I switched to a Mutilate build and equipped some daggers (as seen in the sheet currently). I then set to work on the line trying to implement finisher weighting according to the probabilities in the table. Without going into too much detail, compare Rows 67 and 142 to see what I changed compared to what was there before.

Unfortunately, when I was finished, I had a cycle that seems to make no sense. Because of the insane combo point generation of Mutilate, I used a lot less attacks to do my two finishers, but my finisher damage output was still higher (cell D142). However, due to using less instants, the number in cell J142 is lower. (I don't really understand what this number is...) Also, due to having a higher average number of CP, the cycle time is longer (cell M142). Of most important, the efficiency column (cell N142) is telling me (if I understand it right) that over the course of that cycle 51% of my energy isn't being calculated as used!

I don't understand how the cycle calculation is working, but it definitely doesn't like what I did to it. Unfortunately, I don't have time right at the moment to dig into it.

Basically, all this to say... we are closer but we are not there yet.

I think you found a bug. Expertise isn't applying correctly with mixed weapon types for Humans. I'm going to have to code in separate Main-Hand and Off-Hand Expertise. Looks like its currently only looking at the Main-Hand in order to determine whether Human Racial Expertise is on or not. So, Humans who are using mixed weapon types, where 1 of the 2 weapons is a Sword or Mace, and the other is a Dagger or Fist will have this issue. Granted, that won't be many people, but it will be fixed in next version. Thank you for noticing this.

The problem is that to use the existing combat cycles, you will need to create a 2s/4r/4e cycle or something like that and do exactly what you did. What's happening is that as a combat cycle, it's assuming you go 2s/4r/2s/4r... so you are losing a ton of energy as the cycle assumes you are waiting until SnD drops.

There actually seems to be more than one thing happening here. Let me first point out a few things:

1. Fist specialization (and all other weapon specializations) work on a per-hand basis. If you do not have a fist weapon equipped in the offhand, your offhand attacks will not gain 5% crit chance, even if you have a fist weapon equipped in the mainhand. Therefore, if you swap your offhand from a fist weapon to a sword of equal stats, I would expect to see a DPS decrease, assuming you don't also have sword specialization. At any given time, the character sheet only displays your mainhand crit chance.
   
2. As of 2.4, expertise will also work on a per-hand basis. Therefore, as a human, if you equip a fist weapon in your mainhand and a sword in your offhand, you should expect to see 5% on your mainhand attacks only, and 5 expertise on your offhand attacks only.
   
3. Pre-2.4, it seems that the human expertise racial possibly affects both hands even if only one hand has a mace/sword equipped. Furthermore, equipping a mace in one hand and a sword in the other may be awarding 10 expertise to both hands. However, this isn't really very well tested and may not even be true. In any case, it definitely will not behave this way in 2.4, so there's no need to test it in length.

Sorry if I got a little wordy, but I got the impression from your post that you believe an offhand sword would gain 5% crit from fist specialization, but this isn't true. I just wanted to lay out exactly how the game behaves. Post-2.4, equipping a sword offhand without sword specialization will definitely not be a DPS increase. It may or may not be pre-2.4, for the reasons explained above.

Now, as DMM already mentioned, the spreadsheet isn't handling all of these properly either. I have no idea if the spreadsheet calculates crit chance independently for each hand, but my gut is that it does, because it's been known for quite a while that weapon specializations only affect the applicable hand and do not cross over. DMM already mentioned the bug dealing with expertise.

I hope this clears things up instead of further confusing you.

2.4.0.6 appears to have an issue, i imported 2.4.0.5 into it, and the DPS cells are showing "#VALUE!".

Here's the file : RogueDPS_2_4_0_6NA.xls
Hopefully you can find the issue.

Thanks for the clarification, Drumbum. I wasn't sure, as the in-game display seems to only show mainhand crit. IE if I equip a fist weapon offhand I see no increase in crit chance... so I got the impression Fist spec did nothing for an equipped offhand.

And glad I could help contribute a little bit by pointing out a bug, as well.