We actually do have a very good idea of how the various stats work, backed by a lot of testing. Saying that Recap and Recount produce different numbers is irrelevant: are they measuring different things? are the same combat log settings being used? how many people in the group are running each mod?
With individual testing and combat log data you can accurately test the effects of lots of different game mechanics....if you'd like to know more about that testing is done feel free to browse the various rogue theorycrafting threads or ask specific questions, but suffice to say the spreadsheets have a pretty darn good picture of what's going on: it's not simply hand-waving and voodoo.
See, I think that's you're missing what's meant by something being a 5 % DPS increase. That doesn't mean you're hitting 5% harder necessarily, it means that at the end of the day after counting the increase in damage per hit, the change in crit rate, the change in hit rate, and all the other factors involved, you're voerall daamge output is 5% higher over the same period of time. Your damage output isn't some function of your DPS, it *is* your DPS. So if an item is a DPS increase, that means that it's an increase after taking account all the various factors going into damage *including* hit.
In practice, hit is one of the more valuable stats around so picking up hit tends to be a good move in practice, but in theory there *is* no minimum hit, not even at the specials hit cap. Hit becomes worth even more at that point, but fundamentally it still has some quantifiable worth, so that if you gained a large enough amount of AP by the trade-off it'd be worth falling below that cap: the hit caps don't have some magic property that if you're below them your DPS plummets regardless of the other stats...DPS is a function of all your stats. Hit tends to be one of the more valuable inputs, but there's no magic reason for it, and it some other item produces a DPS increase despite a reduction in hit, it's the better item.
Good point. You successfully pointed out a flaw in my argument. I think the other flaw is my inability to get my point out without rambling and losing my point all together. But that statement set me back.
I had stated in other posts, I think the root of my issues is I just can't seem to bring myself to rely on the number 100% in the sheet that others seem able to do. But that not really an argument in my favor. Just an observation. Thanks for the reply.
We actually do have a very good idea of how the various stats work, backed by a lot of testing. Saying that Recap and Recount produce different numbers is irrelevant: are they measuring different things? are the same combat log settings being used? how many people in the group are running each mod?
With individual testing and combat log data you can accurately test the effects of lots of different game mechanics....if you'd like to know more about that testing is done feel free to browse the various rogue theorycrafting threads or ask specific questions, but suffice to say the spreadsheets have a pretty darn good picture of what's going on: it's not simply hand-waving and voodoo.
Haha, not trying to put peoples hard work down. I have read the theorycraft threads. Some incredible stuff there. As far as Recount and Recap, those where just examples of the two different authors addons reacting differently to the same action. I'm not referring to other people stats so the question of how many other have it (assuming this question was fro syncing with them) isn't relevant. I'm just looking at my own stats. Misses on white hits for example. DPS is another. Since it's just me, both application have access to exactly the same info at all times. My GM uses a Combat Log parser that he posts to a web site that crunches the numbers. This is yet a 3rd set of numbers that don't match up with the two in game mods.
But my point was that you cannot be 100%. But you can be damn close with this sheet and the hard work put into it.
But you all have proven me wrong in my original assumptions and I hate to wonder off into another direction of arguing that may flare up some bad feelings based on my opinion. Let me leave off with, You're right, I'm wrong.
The reason a lot of those numbers are different is because they make different assumptions. The WWS parses your GM uses calculates dps as a function of damage done over time active. Whereas recount uses actual time in combat if i recall. I think recap uses time from first hit to out of combat. So by using 3 different time factors, you will get 3 very different dps numbers for the same set of data. As for misses, some of the addons clump parries/dodges up as a miss and dont separate them. While others actually do treat them differently. The discrepancies you notice to your own stats are probably nothing more than formatting differences than actual differences in the collected data.
Put a crit amount on the unmodeled trinket. Equip that trinket. (I'm using 45 for this example because I want Medallion of the Horde) You will now gain hit instead of crit. This bug works all the way down the line. I have to put my 45 res in the AP column for it to show up in res
Put a crit amount on the unmodeled trinket. Equip that trinket. (I'm using 45 for this example because I want Medallion of the Horde) You will now gain hit instead of crit. This bug works all the way down the line. I have to put my 45 res in the AP column for it to show up in res
The columns for the Unmodeled items have a different header than normal items. You need to look at those headers and not the normal items. That said, I'll probably adjust the Unmodeled columns so they match up to avoid this problem in the future.
The beta version has been pretty stable, hence it's time for a new official version.
2.4.2.2 to be posted shortly...
2.4.2.2 dmm/sp00n Added
Defense to the (Unmodeled) items section.
Unused talent point counter.
Integrated talents, defense and boss level into parry rate.
Integrated defense into miss rate and dodge chance.
Defense rating conversion.
Dodge details.
Mob option attacking from front/back.
Cloak of Unforgivable Sin
Defense to Chest enchant
Delicate Eternium Ring
Elementium Band of the Sentry
Glyph of the Defender
Greater Inscription of Warding
Hand of the Deceiver
Major Defense to Bracers enchant
Moroes' Lucky Pocket Watch
Nimble-foot treads
Slikk's Cloak of Placation
Shard of Azzinoth proc effect (as 45 DPS)
Fixed
Cold Blood for Instants
Helm list to include Wastewalker Helm and (Unmodeled) when all items options checked.
Issue with DPS details difference (was reporting difference with opposite sign).
Removal of more unbuffed references, hopefully resolving open Office issues.
Ghostly Strike option no longer increases damage done.
Buffed DPS no longer shows average Eviscerate damage if no Eviscerates are used.
Gem upgrades are again affected by metagem requirements.
Correct item ID for Figurine Shadowsong Panther
Elixir of Demonslaying now correctly only applies damage to Demons.
Importation issue with +4 stats enchant to bracers.
Importation issue with Heavy Knothide Armor on boots.
Great tool, thanks for all your hard work. I downloaded spreadsheet twice today due to the gemming tool not providing blue gems to select from in dropdown. If I import current build, gems are shown, or if I select blue gem the bonus is included and manually adding gem stat to row provides proper calculations. There simply is no list to select from after doing an item change. I am using an older version (2002) of Excel.
On the Gear/Buffs sheet, check the box labeled "Show non-DPS items" in the Equip Options box (right side of the sheet). Edit: Doesn't seem to work for me, although I believe that's how you're supposed to show blue gems.
I apologize first for my inability to get my download to properly work in my excel. For some reason, I am having issues.
But given that. I would like someone to do me a quick favor and tell me where I should spend my badges.
Currently, I am a comabt dagger specc'ed dwarf rogue (I know, I know. Its just what I like to play).
Right now. I have mongoose on both weapons (and will put that enchant on any subsequent weapons).
I have Riftmaker in MH, and S1 Shiv in OH.
I have 45 badges right now and was thinking of upgrading my OH to the badge dagger (103/dps). But i don't know if that is a clear upgrade or should I wait till I can afforcd via badges the MH dagger. (91/dps --> 103/dps :: not including the difference in equip).
As a side note, I have plenty of honour built up so I can grab the S1 MH dagger too. Just waiting (to see how S2 comes out for honour only or not).
Anyways, just curious on what is the best course of action.
I apologize is this is a repeat question, but it was never really answered: This sheet shows Blade of Infamy/Talon and any offhand better than Akil'zon's as an upgrade over the Vanir set. I take it this is some kind of an error? This is the only sheet that says this. Others (namely the Gear sheet) shows it as a downgrade. and a fairly large one.
Did you try to adjust the build to have fist spec before you checked?
This. In particular, realize that the gear sheet comes with all 5 points in all weapon specs enabled by default, so you must adjust the poison talents by hand for proper comparison of hybrid/non-hybrid specs. The DPS sheet comes with a full talent point tree and present builds, which means that when you swap between various weapons you must adjust the weapon spec properly.
Firstly, I am not talking about hybrid specs. Just simple Fist/Fist vs. Sword/Sword. Also, yes I changed the spec to mimic a typical 20/41 sword or fist spec (with vile poisons) on both sheets. I have been using the sheets for quite a while (this issue was first brought up on page 30 or 31 or something.) Yes, both sheets are also set to Night Elf.
In my gear on the gear sheet:
With the Vanir set I have 1597.58 DPS (with a 21/40 fist spec and poisons of course)
With the Blade Of Infamy + Akil'zon's I have 1583.71 DPS (with a 21/40 sword spec and poisons of course)
On DMM/sp00n's sheet I have:
With the Vanir set I have 1628.09 DPS (with a 21/40 fist spec and poisons of course)
With the Blade Of Infamy + Akil'zon's I have 1630.85 DPS (with a 21/40 sword spec and poisons of course)
Obviously, the point isn't that the numbers differ its that on the gear sheet the change sends my DPS down by over 10 DPS but with THIS sheet it goes up (by about 2, but again, not the point).
One sheet says its a noticeable downgrade, the other that its a justifiable upgrade (I have the vanir set and just got blade of infamy since it was going to be rotted, and have had akil'zon for a while.) I am wondering why this is.
From the outset it looks as though Vanir's set should be better since the item budget from that is both high, and fairly well spent. (The AEP of the stats on Vanir MH is 67.1 and on Infamy 52.64 before the spec talent bonus and damage range/dps) And this makes me more inclined to believe the Gear set (Despite the sword spec talent). But I simply like swords better.
I just want to know what the sheets SHOULD be telling me. Is the Gear sheet right in making it a down grade? or is this sheet right?
EDIT: after playing around with it a bit I am noticing that on the Gear Sheet The fists always wins out and on this one Infamy does. Vanir/Vengful (fist/sword 16/46/0) gives more DPS than Infamy/Vengeful (sword/sword 20/41) on the Aldriana's sheet but less DPS on this sheet. This sheet seems to favour swords and the sword spec bonus more than Aldriana's.
I was just wondering if the spreadsheet took into account the racial expertise bonus that human rogues get using swords/maces? +5 expertise I believe. I just noticed that zero was showing up on the Human row under expertise column.
Keep up the good work
I do notice in the RogueCalc project that Aldriana mentioned that it's showing about a 1% increase in DPS and that much of this comes from better WF modeling than the Gear spreadsheet. This brings up the possibility that the Gear sheet might be undervaluing the synergy between sword specialization and windfury currently. This spreadsheet models the combination as windfury being able to proc from a sword spec extra attack and visa versa. WF cannot proc from WF and SwS cannot proc from SwS but one could have WF -> SwS -> WF chains or SwS -> WF -> SwS chains. I vaguely recall seeing a combat log snippet in one of the threads that seemed to indicate that the attacks could chain like this. It is difficult to prove under normal circumstances as its hard to associate one particular proc to its source but with some well-placed misses and dodges, this snippet showed such a chain. So it might well be possible that this spreadsheet was ahead of the game with regards to valuing Fist spec vs. Swords.
As to Human expertise, I was forced to apply that in a different area as you only get that expertise with specific weapon types. So the 0 shows up in the column because if you are using Fists, Humans get no bonuses to expertise. The extra expertise is applied on the Buffed DPS sheet as a special case for each hand. This was necessary because with hybrid specs, you might only gain the expertise benefit on one hand.
Regarding Fist/Fist vs. Sword/Sword: Another possibility is that there are slightly differing (de)buff sets between the two sheets. And yet a third possibility is that it's simply in the modeling differences. The margin of error on both sheets is somewhere at or above 1%, and when you are looking at 1600 DPS, then 10 DPS (or 2, for that matter) is well within the margin of error. Basically, at that point, I guess you choose whichever spec you trust/like better.
Using the 2.4.2.2 sheet (as well as at least the one preceding beta), I am unable to save new talent builds. I get all the pop-ups for saving, type in a name for the new build, but no new entry gets added to the dropdown list, and a different build gets selected and loaded.
I'm working on an expose weakness calculator and I'm trying to get a good number to use for the average amount of dps a raiding rogue would get from AP. Right now I'm using 0.351 dps per AP which is the amount I got off of the spreadsheet with my rogue's gear (mostly T5 level), but I'm wondering if that's a good average value to work with or what would be a reasonable number.
This spreadsheet models the combination as windfury being able to proc from a sword spec extra attack and visa versa. WF cannot proc from WF and SwS cannot proc from SwS but one could have WF -> SwS -> WF chains or SwS -> WF -> SwS chains. I vaguely recall seeing a combat log snippet in one of the threads that seemed to indicate that the attacks could chain like this. It is difficult to prove under normal circumstances as its hard to associate one particular proc to its source but with some well-placed misses and dodges, this snippet showed such a chain. So it might well be possible that this spreadsheet was ahead of the game with regards to valuing Fist spec vs. Swords.
Our best current knowledge is that white attacks can proc either sword spec or Windfury; Windfury attacks can proc only sword spec; sword spec can proc no other extra attacks. I don't recall the last time this was experimentally verified, so I'll see if I can drag a shaman out to Blasted Lands at some point to see.
If that's how the gear sheet models it, that could very well explain the difference between the two. If sword spec can't proc WF, then the white DPS model will fall off a bit less than 1% or so (~20% loss to ~5% of white MH DPS?). At white DPS being roughly 60% of the total, 1% of 60% of 1600 = 9.6, so we're in the right ballpark. Of course, the exact number is likely to be quite different from my incredibly rough napkin math. I just wanted to establish that we were in the right magnitude range.
This is what I am most interested in. In fact, what I really want to see is a good method for adapting to and modeling variability within cycles based on a ruleset. This is something that I really feel isn't adequately captured right now in the DPS sheet with regard to Mutilate cycles, and I'm the one who wrote those cycles. After all, there is a big difference between the cycle time for using 1 mutilate and using 2 mutilates, and saying it takes "1.5" mutilates to achieve the average cycle is... awkward. So even though I think we have a decent model, I really think it would be a lot better in a program form, and I'd be very interested to see what you have.
So, I wanted to bring this discussion to this thread real quick because you've already got a good framework laid out. Currently your probability tables provide the average number of CP generators performed, given a "target" number of CP (X and Y for two finisher cycles), and the probability of reaching X, or X+1, or X+2 (or the same for Y). But you can extend this, and also compute the average number of CP generators performed to reach X, X+1, or X+2 (or the same for Y). Now what you know is that any given iteration of the specified cycle can either take the form X/Y, X+1/Y, X+2/Y, X/Y+1, etc., up to nine combinations of those. And your existing cycle models already compute what any of those individual cycles looks like -- plus you know roughly how many CP generators you're spending specifically for each of those individual possible cycle iterations. So instead of patching average finisher effects and CP generators into the existing cycle models, you can take the weighted average of each of those possible cycle iterations.
Now take this a step further; since there's a big difference between Mutilating and not Mutilating (up to 3 combo points and 60 energy), it's not entirely accurate to say "it takes, on average, 1.4 CP generators to reach X number of CP." But using your existing probability tables, instead of computing the average number of CP generators to reach each of X, X+1, and X+2, you can take each possible combination of reaching X/X+1/X+2 and using N={0,1,2,3,4,5} CP generators. Then, not only do you average each of the cycles X/Y, X+1/Y, X+2/Y, X/Y+1, etc., you also duplicate those cycles for each of the possible number of CP generators it can take to reach X/Y, X+1/Y, etc., so that you are accounting for each individual possible cycle. Assuming a 3+s/5r Mutilate cycle, this means you need to calculate:
- Each of the cycles 3s/5r, 4s/5r, 5s/5r, times...
- Each of the probabilities of hitting those CP targets, times...
- Each of the possible combinations of CP generators used to get to each of those cycles, times...
- Each of the probabilities of requiring those numbers of CP generators to get to each of those cycles.
Is it complicated? Yes. Is it possible, given your existing framework? Yes.
I have a decent working model that incorporates separating the cycle out into an average of each possible X/Y, X+1/Y, etc., but it does not incorporate separating it into an average of each possible number of CP generators used, as I did not consider that potential degree of accuracy when making it. It's really a question of how much accuracy you need. We've modeled cycles using an inline average of Ruthlessness and Relentless Strikes effects for quite a long time, and I found that when I implemented an average of each possible combination of those things that it didn't make a very big difference at all. While Mutilate is probably a bit bigger of a difference, I'm not sure if it's enough to be worth the pain it would take to implement it with 100% accuracy in a spreadsheet.
I believe weighed damage based on probabilities is what is occurring now in the cycle sheets and that the 1.4's for number of instants is a weighed average. Now, I can conceive of some value in expanding out the cases, but I keep getting stuck on how to parse these values into something useful. The energy queuing and capping issues when one gets an unlucky streak have effects but the difficult part is how to determine how often this actually occurs and modifying the damage accordingly.
The nice thing about the existing framework is the potential application of new types of cycles alongside the xs/xr cycles. One really could model a 1 instant then Snd/2 instant then Rupture cycle and compare this to 3+s/5+r cycles. Could label it as 1[i]s/2[i]r or something. It's a direction I've wanted to go, but I'm not finding too much time lately to work on serious additions to the spreadsheet. To some extent, I'm saving up some mental energy to address the impending new rogue talent announcement.
Obviously, I'd rather have the spreadsheet represent the correct mechanics, so if you do run across more evidence showing that the interaction acts differently, I'm certainly interested. The existing framework isn't that difficult to modify in terms of what can proc off things and what cannot proc off others. For now, I'm still going with my best at least semi-educated hunch that WF and Sword Spec can proc off each other but not themselves.
If that's how the gear sheet models it, that could very well explain the difference between the two. If sword spec can't proc WF, then the white DPS model will fall off a bit less than 1% or so (~20% loss to ~5% of white MH DPS?). At white DPS being roughly 60% of the total, 1% of 60% of 1600 = 9.6, so we're in the right ballpark. Of course, the exact number is likely to be quite different from my incredibly rough napkin math. I just wanted to establish that we were in the right magnitude range.
Ah, so it should be it is.
Then my question (which I hope Vulajin doesn't rip on me for) switches focus: Which is/should be better. I am going to test it myself and if there is no discernible difference between the 2 I will stick with the sowrds.
But which SHOULD be better? For a night elf would the sword spec + Blade of Infamy/Akil'zon's outweigh fist spec + Vanir's fists? Vanir's fists are surely better items than Blade of infamy (by a bit) and Akil'zon's (by a lot) and the spreadsheet leave me with no final answer.
Then my question (which I hope Vulajin doesn't rip on me for) switches focus: Which is/should be better. I am going to test it myself and if there is no discernible difference between the 2 I will stick with the sowrds.
But which SHOULD be better? For a night elf would the sword spec + Blade of Infamy/Akil'zon's outweigh fist spec + Vanir's fists? Vanir's fists are surely better items than Blade of infamy (by a bit) and Akil'zon's (by a lot) and the spreadsheet leave me with no final answer.
I think from these findings you can assume they are incredibly close to each other (either a bit above or below). Sure, the fists have some nice stat allocation, but inherently sword spec is better than fist (extra MH attack > extra crit %), which makes up any difference in stats. In the end, how you use your weapons plays so much more into their effectiveness that a few stat based DPS either way doesn't really matter. At this point go with the one you identify most with. They are both excellent choices.
06/09/08, 1:55 PM
Similar Threads
