I understand that, so I'll qualify my statements.

How much armor before exec applied are you basing your figures on? Can we all agree that there is probably a general raid makeup that would suit most raids and thus give a general/average boss armor amount? If this is a yes, as I suspect/hope it will be, then giving a figure off those numbers would suffice for now. Again, I realize that the proc is still an issue, but until its resolved, we've all agreed to use the 1 PPM so that's a non issue.

Of course the exact number will be very complex for each person, but if we can assume or at least agree partially upon a general normal boss armor after raid debuffs and then move from there with both enchants/procs and see a general idea of %dps increase.

The point I'm making is, if you assumed the same armor as ald, then it would seem your math differs from his and shows exec being ~2% more dps than Mongoose. If you assumed different, then we can look at a redo of the math with the inclusion of different armor values and go from there.

It seems like we're making this much harder than it has to be about getting perfectly exact numbers, when in fact, you've all already stated that getting exact numbers for each person is nigh impossible. Every person's gear affects their relative value of each enchant differently. So if we take a normal/base raid boss armor value, minus "normal" raid buffs and then include the effect of each enchant, can't we get a general amount? And this only comes up because I see 3-4 different sets of numbers, one being the spreadsheet I know and love, and would like to know which set of numbers to base my assumptions off of.

The armour values I used are the 4 digit numbers at the start of each line, I used the same debuffed boss armour as Carnivori. My post was only made to show that the numbers posted above were incorrect. Doing "manual" calculations on this subject is totally pointless, the two enchants are too close to decide between them based on rough numbers. Even when the new sheet comes out you may find that it depends on personal factors such as arp gear etc. One cannot claim that one is better than the other in general solely based on napkinmath, anyone doing so should be regarded with a great deal of suspicion.

I posted a ~9100 swing combat log of double mongoose enchants with same speed weapons on autoattack. It seems to of been completely overlooked but could help confirm the proc rate on mongoose for comparison's sake. Page 11 post 271 of the mongoose vs executioner thread. or just click this: Free file hosting by Savefile.com

I also included madness of the betrayer, exalted hyjal ring, and tsunami talisman. Just cause it couldn't hurt to double check it against currently used proc rates/cooldowns.

I think that this is a point that is often ignored in this debate and is something that will be reflected in a spreadsheet. The fundamental problem, as I see it, is not knowing the exact proc behavior so that it can be correctly modelled. The sheet will then help people understand what works (and how it works) for them based on their gear, and will probably include a lot of synergistic behavior that we fail to calculate on the back of an envelope.

Just as a concern, with the new expertise rating that was introduced in 2.3, how should this work with hit rating? I'm aware that the cap is 89 for the rating (for a total of ~22.4 expertise), but I'm still looking into ways to merge the two to work well with each other. Any input from the community would be greatly appreciated, as at cap one can actually stand at any angle from a mob, unless it cleaves, and not have to worry about parry (89 expertise pushes dodge and parry off the table apparently), thus the tank should not have to worry about taking extra damage due to a parried attack.

As far as I know we only have estimates of the average boss dodge/parry rates. In addition, just as Void Reaver has high armor, various bosses will have high or low dodge or parry rates. In addition many bosses block, and expertise does not negate the chance of a block, so you will still want to stand behind.

What do you mean by merge the two to work well with eachother? They serve similiar purposes, however there is no magic number or ratio that will be the best. There is nothing special about being at the hit cap or the expertise cap, they should just be kept in mind so that you don't go over them and not get the benifit from the stats. What you should be looking at is which piece of gear is more valuable and not which one will increase my expertise/hit rating the most.

This is a very long post to ask a very simple question about gear that you should be able to answer for yourself using the spreadsheet.

Yes, that is true. However the spreadsheet will only go so far. I need to wait for the updated version that includes some of the pieces that i am wearing before I can tell. My lack of any type of spreadsheet editor at my current location hinders my ability to look at things, let alone change them. I'm in the process of number crunching at the moment to get some rough estimates, but even a small "yes" or "no" from the general rogue community would be greatly appreciated. I do thank those of you who look at this thread and its posts for your time, as you have created a beautiful spreadsheet that clearly beats the other one on accuracy, and I do look forward to helping out with it once I get the time to sit down and look at what goes where. I have not found much about how expertise falls in with melee, other than that it reduces the dodge and parry chanced of a mob. how this will improve DPS on the other hand is something that I have not seen. If I had the ability to create a thread on it I would, but the system tells me "no" every time I try.

Dodge appears to be fixed at 6.25%. There was a post somewhere where a guy had 25 expertise and hadn't been dodged in SSC/TK in 3 weeks. Thats not to say other bosses are tuned differently, but its likely a static 6.25% dodge rate. Parry seems to fluctuate a bit and best estimates put parry rate on bosses anywhere from 10 to 15%. I would NEVER suggest being in front of the boss unless absolutely necessary.

And yea.. use the damn spreadsheet.

So, long story short:

On white attacks, Expertise rating is point-for-point virtually identical to hit rating. However, it does additionally contribute slightly to yellow attacks; hence, depending on gear, Expertise Rating contributes damage roughly equivalent to 1.1 to 1.2 points of Hit Rating. For instance, The Vashj belt can reasonably be modeled as having about 27 hit rating instead of the expertise for purposes of estimations

Thanks for that Aldriana, I'll note that for use later on. And I do use the spreadsheet, I just needed some explanation to how expertise related to hit. And to you Aldriana, it is worth it trying to go for expertise over hit? Or is it just something to look at as a compliment to the hit rating? I know the skill is different for tanks, just trying to strap a rogue perspective to it to see its use, since the combat tree gives us 10 points in it anyway.

Thanks for the advice. I fell that I have received enough information to go in the correct direction. Thanks for the input, and I'll try to sit down with the spreadsheet later on, seeing that it is an off night from raiding.

The following questions are concerned only with a combat rogue's boss damage from behind.

In my never-ending search for truth about what exactly my "attack table" looks like from behind during raids against various bosses, a few questions have emerged:

1. How does expertise rating affect yellow damage?

Until reading Adriana's last post, my understanding of expertise rating was this:
For melee auto-attacks from behind, weapon expertise replaces dodge chance with hit chance on the attack table. This makes expertise a statistic with its very own cap, completely seperate from the hit cap, as hit rating replaces miss chance with hit chance. Thus, even if I have reached the hit rating cap, I can still increase my chance to hit by raising my weapon expertise. However, the attack table governing yellow damage remains a mystery...or does it?

2. As Latito noted above, my chance to be dodged from behind is approximately 6.25%. According to Wowwiki, it's 5.6%.
Copied from Wowwiki:

Since a boss-mob is calculated as level 73 mob its dodge & parry is increased by 3*0.2% = 0.6% up to 5.6%. To negate 5.6% dodge & parry 22.4 points of expertise or 88.256 points of expertise rating are needed.
So the current expertise-cap against boss-mobs is at 89 points of expertise rating.

For a combat rogue swinging from behind, what are the real numbers?

3. Apparently, a 73 mob cannot block or parry attacks from behind, but it can still dodge them. If so, if I move behind the 73 mob, are block and parry now converted to hit chance on the attack table?

How do you delete a message you accidentally posted twice? I can't seem to get rid of it.

Yellow attacks are on a two-roll system, one to determine if it lands, and then another that determines if it crits or not. That's why Expertise isn't as dumb as we would have thought it was when Surprise Attacks was touted as a DPS loss (in a re-evaluation, knowing what we do now, it probably would have been ... "ok" - the current form is far from broken as a 41 point talent, however.)

Hi,

on a totally unrelated matter:

Is there any discussion on Loot-Priority for Cursed Vision? We have Illidan on farm for 2 months now. Our guild-Leader has decided to give this to Rogues first (Usually we just auction all Items with a standard DKP System). 2 have it, 2 still need it. Our Hunters are making a major deal out of this. I mean MAJOR deal.

Can anyone provide me with a link to a discussion or any good arguments to shut those Hunters up once and for all? Or do they actually have a point?

All Hunters have 5/5 T6 already. Rogue tokens still are pretty high in demand.

Oh and sorry for the bad english.

So, I was tinkering with some math earlier on armor penetration, because it has come up several times that it's not exactly accurate to say that an armor penetration buff of P with uptime of U provides UP armor penetration on average. To start with, it's necessary to know that damage reduction from armor is determined by:

DR = A / (A + F)

Where DR indicates the percentage of damage reduced, A indicates the defender's armor, and F indicates a factor based on the attacker's level (F = 10557.5 for a level 70 attacker). Trivally, we can subtract this from one and get a formula for the fraction of damage that gets through the defender's armor:

DMG(0) = F / (A + F)

Now we want to find out how much damage gets through when the attacker has P armor penetration. Trivially, this can be shown to be:

DMG(P) = F / (A + F - P)

Now we want to figure out the average damage that gets through if P is provided by a debuff with uptime U:

(1 - U) * DMG(0) + U * DMG(P) =
(1 - U) * F / (A + F) + U * F / (A + F - P) =
(F - FU)(A + F - P) / (A + F)(A + F - P) + UF(A + F) / (A + F)(A + F - P) =
(FA + F^2 - FP - FUA - F^2U + FUP +UFA + UF^2) / (A + F)(A + F - P) =
(FA + F^2 - FP + FUP) / (A + F)(A + F - P) =
F(A + F - P + UP) / (A + F)(A + F - P)

Now that we know how much damage gets through with that debuff, we'd like to come up with a static value, X, such that a constant armor penetration of debuff X provides the same damage penetration as effect P with uptime U. We can do that by solving this equality:

F / (A + F - X) = F(A + F - P + UP) / (A + F)(A + F - P)
(A + F - X) / F = (A + F)(A + F - P) / (F(A + F - P + UP))
A + F - X = (A + F)(A + F - P) / (A + F - P + UP)

A + F - (A + F)(A + F - P) / (A + F - P + UP) = X

This is our final formula for X. Now, there are two questions about this formula:

1) Is it commutative, i.e. does the order in which the debuffs P(n) with uptime U(n) are applied matter?
2) How different is this result than the one granted by simply assuming X = UP?

The answer to the first question, very briefly, is no. A full mathematical proof of this fact is not necessary, because you can simply run the formula yourself using, for example, P(1) = 840, U(1) = 0.45, P(2) = 1000, and U(2) = 0.35 (these being the example values I used) and observe that applying effect P(1) first results in a very slightly different overall result than applying P(2) first. However, the difference is minsicule (less than 1 armor), leading me to believe that this formula can be used commutatively for the purpose of DPS modeling with extremely minimal negative effects.

Regarding the section question, I found that most of the values X provided by this formula were within 3-4% of the values provided by simply letting X = UP. For a single armor penetration proc (say, Executioner), the difference will be minimal enough that it shouldn't matter too much. However, for several different armor penetration procs (say, if you used Executioner, [Warp-Spring Coil], and [Madness of the Betrayer] all together), the individual small errors may add up to an overall noticeable error.

Since armor penetration effects with less than 100% uptime (i.e. procs) are not exactly plentiful, and many of them ([Icon of Unyielding Courage], [The Night Blade]) are not optimal for most rogues, I'd say that modeling armor penetration procs as X = UP is perfectly fine for now. If armor penetration procs become more prevalent in the future, however, it may be advantageous to switch to the formula above.

So far, since Madness is a fairly small armor reduction with high uptime, I've just been approximating it as a static effect with armor penetration 300 * uptime, which I think is probably pretty close to accurate. This means that until the last patch there was only one significant limited-uptime proc to worry about, that being WSC, so I just modeled out the two cases and summed them explicitly - more or less as you did above. With Executioner, my initial inclination is actually to work out all 4 cases, for two reasons:

1) 4 cases isn't very many, so it's managable in scope.
2) It allows one to factor in the Armor Pen cap

If there start to be more large armor pen procs, I might have to reconsider this approach, in which case Vulajin's approximation is probably the way to go for a spreadsheet. Ultimately, of course, optimal is to use a programming language to recurse over all the procs and sum the average directly, which is clearly the right way to do it but takes us back to the "doing this in a programming language would be a heck of a lot of work, and I'm disinclined to do it all myself" point.

Finally got around to parsing this data set (and, in case you were wondering, yes, this is part of the final tweaks to the spreadsheet before 0.9 - expect it tonight or tomorrow - basically as soon as I figure out what proc rate I want to use for Executioner and decide what of the 2.3.2 changes to include), and figured I might as well post the results:

Madness of the Betrayer: Proc rate is between .84 and 1.10 PPM, which matches the currently-used value of 1 PPM.

Band of the Eternal Champion: No procs closer than 59.66 seconds, and a dozen between 60 and 65, meaning a 60 sec internal cooldown. Sorting out eligible attacks from the rest, we find that the proc rate lies between .76 and 1.20 PPM, which jives with the 1 PPM and 60 sec internal cooldown previously measured.

Tsunami Talisman: No procs closer than 45.05 seconds, and about 20 less than 50 seconds, confirming the currently-used 45 sec internal cooldown. Proc rate on eligible crits is between 9.4 and 13.7%, which matches the currently-used figure of 10%.

Mongoose: Well, this one is complicated by the presence of stacking procs from the two weapons, but with a few assumptions and some clever parsing I got some usable numbers out of it. According to my best guess, the proc rate of Mongoose, based on this data set, lies between 1.03 and 1.31 PPM. Now, this just barely misses the widely accepted value of 1 PPM, which means one of three things:

A) This is a flukish set of data, lying in the 5% tail,
B) The stacking procs are causing confusion over proc refreshes which is distorting the data, or
C) The proc rate isn't actually 1 PPM.

It should be noted, for those that might be inclined to lean towards option C, that that range doesn't line up very well with the Executioner proc rates we were deriving, either; best guess I had on that was 1.33 PPM, which is just barely outside the range of this dataset.

So, long story short: we still don't know what's going on.

I know that all currently-known internal cooldowns are longer than the proc length, but that needn't necessarily be the case.

Could there be (say) a 5 second or 10 second internal cooldown on Executioner or Mongoose?

Why do you coin it "stacking procs"?
The standars UI clearly shows two separate bufficons for the mongoose proc.
So there must be a way to distinguish MH and OH procs.

I was under the impression that you miss the "You gain.." message only
a) If the buff in question is a single buff and b) the refresh comes from the same hand.

Correct me if I'm wrong here.

Yes, but that's still a potential source of confusion! There's also a further potential error, which is that (I believe) the "Mongoose fades from you" message is not specific to MH or OH.

For example:

0s You gain Mongoose (MH)
8s You gain Mongoose (OH)
23s Mongoose fades from you (unspecified)
29s Mongoose fades from you (unspecified)


Is that 15s on the OH and 29s on the MH (i.e. three procs, with an MH refresh), or is it 23s on MH and 21s on the OH (i.e. four procs, with MH and OH refreshes)?