Shortest spacing: 44.985 seconds.
Procs: 51
Crits outside 45 sec cooldown: 506
Avg Proc Rate: 10.1%
Std Dev: 1.3%

So... 45 sec internal cooldown, 10% proc rate it is?

I think with the advent of different dmgreduction effects and the nonlinearity
of the reduction formula the extra computation for WSC-reduction should go away.
IMO it should be the same as it is done for Madness of the Betrayer.
(thottbot e.g lists the effect of Madness, WSC and T6 items as the same spell)

In the spreadsheet there is an error regarding AP critproc trinkets.
In the Trinket sheet all the AP increasesare summed together and this sum is then multiplied with the uptime computed for the tsunami talisman.

That need not be correct for the Hourglass or other trinktes that may be released.

The extra step is done precisely because armor penetration is nonlinear; WSC is modeled as a 1000 armor reduction that's up, say, 25% of the time, rather than a 250 armor reduction that's up 100% of the time, as that more accurately reflects reality. In truth, Madness of the Betrayer would be more accurately modeled as 300 penetration that's up ~50% of the time rather than ~150 armor penetration. I do not do so for two reasons:

a) This becomes a very messy calculation in very short order if you compute every armor penetration proc seperately, and
b) As it's a smaller armor decrease, the nonlinearity has less of an effect.

Yes, all of them actually have the same effect - they reduce armor. And the sheet models them all the same way - they reduce armor.

1) It *is* correct for Hourglass; data collected indicated to my satisfaction that TT and Hourglass exhibit the same proc behavior.
2) Critproc AP is defined as "AP from procs that work the way TT and Hourglass do". So any trinket in the future that procs by a different pattern wouldn't use that column, because it wouldn't be "critproc AP" the way I've defined it.

So, couple of things:

1) I was doing some testing this evening on live, and as an incidental benefit of what I was *actually* testing, took some data on Tsunami Talisman. Out of 59 total procs, the spacing never was under 46.281 seconds. So the values on test would appear to not be changed - the spreadsheet was just using slightly wrong values. I'll add it to the list of things to fix.

2) What I was *actually* testing is the proc rate of 4/5 T5. Offhand, the answer seems to be "about .5 PPM with no internal cooldown" - I'll need to take a larger data set at some point to confirm that, but that's my best guess at this stage. If that is, in fact, true, that means that the spreadsheet is currently overvaluing that set bonus somewhat for sword rogues and by quite a bit for dagger rogues. Again, added to the list, pending confirmation.

3) Re: Trash DPS scenario. I haven't actually run any numbers on this, so trust the following assertion only assuming you trust my gut instincts:

As a general rule, you want to keep SnD up as much as humanly possible. So if there are additional mobs to be killed on hand (i.e. if you're killing a pack of stuff and there's more stuff left in the pack), I typically refresh the SnD. If, however, there's no further mobs in the pack (or the next DPS target is far away), I'll generally throw the Evis if it's 3 points or larger, *or* if I have energy for it but not a SS (i.e. my current energy is 35-39).

Again, no real theory backing that, that's just my instinct.

Update on 4/5 Proc rate:

With 2x1.3 speed weapons, I had 36 periods of uptime including at least 49 distinct procs out of 4103 attacks. 95% confirence interval: .86% to 1.52%, aka .40 to .70 PPM.

With 2x2.6 speed weapons, I had 57 periods of uptime including at least 69 distinct procs out of 3535 attacks. 95% confidence interval: 1.50% to 2.41%, aka .35 to .55 PPM

So... .5 PPM certainly looks to be the more likely outcome, but 1.5% proc rate cannot be entirely ruled out.

...sigh.

Guess we need, you guessed it, still more data to resolve this one.

So you favor the money belt over Vash'j's belt now? Gratz on the t6 shoulders. :P

just screwin with the spreadsheet, adjusting boss armor to 4000 instead of 5000 puts it more in line with where I was.

Re: Shoulders. Thanks

Re: Belts. Don Alejandro's and Belt of One-Hundred Deaths are extremely close in terms of rating by the spreadsheet (in the current version I'm using, which has about half of the changes for 0.7.8 in it, they score within .2 AP of each other assuming late-BT quality gearing). So, ultimately, it's a matter of personal preference - subject to a couple of assumptions.

First off, it should be noted that the exact behavior of weapon skill on dodge has not been finalized. As such, it's entirely possible that the Vashj belt is under- (or over-) valued in the current incarnation of the sheet. So when and if an answer is ultimately reached for that issue, the answer may change here.

But, for the sake of argument, lets assume, for the moment, that Weapon Skill is modeled correctly. At this point, the decision - for me at least - comes down to two things. Interrupted combat, and rounding.

1) Interrupted combat. The sheet, as has been noted, assumes endless uninterrupted combat. In practice, very few boss fights exhibit uninterrupted combat; for instance, in all of SSC/TK, there are really only two fights that have even the potential for uninterrupted sustained DPS: Tidewalker (assuming no graves) and Void Reaver (depending on your strategy). And, looking through Hyjal/BT, they're not really any better; you have Gorefiend, Bloodboil, Anetheron, Kaz'rogal... and that's about it. Moreover, you'll note that many of *those* fights are fairly short, and thus are poorly modeled by "endless". So when you add it all up, most fights shorter and/or more interrupted than the spreadsheet models.

The question, then, is what effect this has an actual DPS. Well, interrupted fights have a number of effects; the major one, of course, is that your cycles necessarily change. Additionally, it increases the value of effects with cooldowns. But the other fairly fundamental thing it does is that it change the ratio of white to yellow damage; whenever you're in a situation where your energy is regenerating but you're not autoattacking, you're going to do more yellow damage relative to the amount of white damage. This will thus relatively decrease the value of hit, which only affects white damage; hence, between two otherwise comparable items, it is generally superior to wear the one that has less hit on it.

With regards to the specific belt situation:

Don Alejandro's has no inherent hit on it, so will gain only what it gets from gems - most likely 5 or 10 hit rating.

Belt of One-Hundred Deaths has 25 weapons skill rating = 6 weapon skill = .6% hit = ~9.5 hit rating on it inherently, plus whatever additional you gain from socketing (which is almost certainly nonzero). Additionally, some of it's benefit is also from -enemy dodge, which also helps white damage more than yellow.

Hence, Belt of One-Hundred Deaths is more of a white-damage-friendly belt, while Don Alejandro's is more of a yellow-damage-friendly belt; hence, in interrupted scenarios, the advantage tends to go more toward Don Alejandro's.

It might be noted, however, that Heroism - particularly multiple Heroisms per fight - can tip the balance more towards white damage, as can haste pots. As such, there is a reasonable argument to be made for favoring white damage more than yellow. And in case you were wondering, yes, I am planning to add "drink a haste pot every 2 minutes" and "give me N heroisms every 10 minutes" as options to the buff sheet at some point; I just haven't gotten around to it yet.

So, to summarize this point: there is a reasonable case to be made that Don Alejandro's will perform somewhat better in interrupted-damage scenarios than Belt of One-Hundred Deaths. The real answer here, of course, would be to have both belts and rotate depending on the fight. But that's not always possible.

2) Rounding. For reasons of continuity, the sheet doesn't quantize to integers like the game does. As such, for fine-level decisions like comparing these two belts, it may be worth considering your position relative to various roundoff cutoffs. As an example: 25 weapon skill rating is modeled in the sheet as 6.34... weapon skill, when in reality, it's only 6. As such, In terms of actual performance in real-world scenarios, the Vashj belt is overvalued by the spreadsheet by roughly 1/3 of a point of weapon skill, or about 2.3 AP. Throw in the fact that, with end-game optimal gear, Don Alejandro's puts you better positioned relative to the roundoff points of Vitality for both Agi and Sta (trust me, it does), and you'll find that in a real world scenario, Don Alejandro's gains a couple of AP relative to the Vashj belt due to these roundoffs. Again, not a major deal, but for two belts as closely spaced as these, every little bit matters.

In conclusion: it really is a matter of personal preference. For the reasons above (and some other, even more vague ones, like how hard it may or may not be to convince your guild leadership to let you socket a Vashj drop with BT gems relative to a Hyjal drop), I do have a slight inclination towards Don Alejandro's. However, I could easily see someone having the reverse preference; there's really no wrong answer between these two.

Odd; I did an ungraved Tidewalker attempt a month or two ago and found my damage to be within 5 DPS of what the spreadsheet predicted. I will take a look at your data set and the spreadsheet and see what I can come up with.

Well, detailed analysis from a screenshot is hard, but here's my take on the matter:

1) I may be a bit off in terms of what Tidewalker's armor is. This comes back to my desire for a more systematic study of how much armor various bosses have.

2) There will always be a certain amount of variance in these things anyway.

3) As the fight was not endless, it is likely that some of your cooldowns and whatever were able to be used more than is modeled by this sheet (in particular: AR is a 5 min cool, so if you're able to use it, for instance, twice in 7 minutes, that's going to make something of a difference).

4) A member of your group mentions in chat that he missed the heroism. Presumably this means heroism was used, and, as you were never graved, you probably did benefit from it, which would increase your DPS by a nontrivial amount.

Fundamentally, though, even given 1-3 above, I am somewhat concerned about this discrepancy, as it does seem overlarge. However, there's not a lot I can do to debug it just from a screenshot; a combatlog, or WWS, or even a Recap, would be more useful to me.