The Retribution Concordance - 4.3, Voice of the DPS - Page 40

Naididae · 02/02/11, 2:32 PM

Judgement, SoT application and damage proc do not seem to proc "on melee attack" trinkets.
Only CS, TV and auto-attacks proc them on live.
You can easily test this by equipping License to Slay and noticing that you get only 1 stack on an auto-attack, no matter how many stacks of censure you have, or that you get 1+1 stacks if you CS exactly when an auto-attack is landing.

Assuming License to Slay model extends to all trinkets with "on melee attacks" in their proc text this seems to be the expected behavior on such trinkets.

As a personal experience, Heart of Rage seems to proc only from these sources too. Although I didn't do long tests.

chippydip · 02/02/11, 4:55 PM

This got me thinking: How much benefit do we really get from stacking a temporary strength buff with the rest of our cooldowns?

AW obviously provides a +20% boost
Zeal we'll ignore for the moment (since I don't happen to know a good value off the top of my head)
GoAK also provides a +20% str boost when fully stacked

GoAK can't be used every time, but it can be used at least 1/3rd of the time and as much as 50% or more of the time, depending on the specific length of the fight. For example, if the fight is less than 10 minutes so you can't squeeze 3 GoAK uses in, then you can just delay your second use until AW+Zeal are ready. Similarly, if a fight is in the 5-6 minute range so that you can get a second GoAK but can't quite get a 4th AW+Zeal you can delay your last AW+Zeal until GoAK is ready.

Overall, this results in an average boost of 28%-36% from AW and GoAK over typical fight lengths for a "Use" trinket compared to 5%-8% for a static stat. With 1/6th uptime on a 1605 Str trinket, it will be worth 342-363 Str compared to a static 321 Str that would be worth 337-346 Str.

But wait, it gets better. Lets say Zeal is a 1% overall dps boost (ie a 6% with a 1/6th uptime). This raises the "Use" bonus to 35%-44% and the static bonus to 6%-9% giving us an average Str boost of 361-385 vs 340-350.

In fact, that's not even the end of it! Since "Use" items are available right at the start, 1/6th uptime is actually the worst case. Consider a 6 min 40 sec fight where you use the trinket 4 times (with cooldowns). Uptime in this case is actually 1/5th which makes that "Use" trinket equal to 433-462 Str compared to the 343-356 of the static trinket! (The static value gets a slight bump from the increased uptime of AW/Zeal/GoAK over this duration, bringing the buff into the 7%-11% range).

So, even in the worst-case fight length where the trinket comes off cooldown just as the boss dies, a 1605 use trinket seems better than a static 321 str and for other fight lengths its even better.

What about fights where you save cooldowns for particular phases? True you might get less use out of the trinket and cooldowns, but if you're doing this you're already sacrificing overall damage for burst at a specific time and a "Use" trinket helps you do that more effectively.

What about proc trinkets? My guess would be that, averaged over enough fights, the proc won't line up with cooldowns any more than a static effect will. It may be possible to wait on cooldowns for a proc, but once you line up with the proc, its likely it will proc again just before your cooldowns become active making it difficult to time further procs. In addition, its even trickier to try to line them up if you're using GoAK and have to plan AW+Zeal ~10 seconds in advance. Finally, any delay to wait for pocs has the potential to cost you the use of a set of cooldowns if the fight ends at a bad time. Overall it may be possible to benefit from procs more than average, but its certainly not something I'd be interested in focusing on during progression fights and the gains don't seem likely to be anything stellar.

What about engineering? If you have a use trinket plus the enchant you'll want to use the trinket (more str) during cooldowns which means the enchant can't be used then. This could still be a slightly bigger boost to Str than other profs, though, since the enchant benefits from fight length as well and could end up with a bit more than 1/6th uptime.

What about [Fury of Angerforge]? This could either be better than [Impatience of Youth] or worse depending on the specifics of the fight. Fury isn't usable at the start of the fight until it has time to stack up. If you wait for it to stack and then pop cooldowns and end up missing out on a cooldown use due to fight length you've just cost yourself some dps. (If you use Fury without cooldowns then its clearly worse than Impatience with cooldowns.) Fury also suffers if the fight has time where you can't melee for long enough that the stack falls off. If this happens soon before your cooldowns are ready, you're right back in the same situation of having to wait for it to stack or Impatience being a better choice for the fight. On the other hand, if the fight duration and mechanics allow you to use Fury just like a normal Use trinket, then it may actually be our BiS trinket.

(I was actually very surprised when I started thinking about this and running some numbers. I currently use Impatience, but mostly just because its so easy to get and I figured using it with cooldowns should bring it close to a static 321 benefit. The fact that it could actually be worth as much as a 400+ static Str was totally unexpected.)

(edit: of course I meant Zeal when I said TV--fixed)

chippydip · 02/03/11, 2:22 AM

Originally Posted by Naididae

On AW don't you have to consider it actually being a 10% bonus to str? Since it's 20% to damage and each point of str gives ~2 dps, so you have to cut the 20% in half to see the raw str you get in reality.
This should put both types of trinket closer following your logic.

It doesn't matter how much dps you get per point of str (or crit or haste or mastery). When AW is up everything does 20% more damage which makes all stats 20% more valuable.

Lorayth · 02/03/11, 9:30 AM

Originally Posted by chippydip

(I was actually very surprised when I started thinking about this and running some numbers. I currently use Impatience, but mostly just because its so easy to get and I figured using it with cooldowns should bring it close to a static 321 benefit. The fact that it could actually be worth as much as a 400+ static Str was totally unexpected.)

Two problems with your calculations. One, your calculations are based on pie-in-the-sky scenario of when the stars align and angels dance over your head, where the fight length is exactly that you can pop the on-use trinket when the fight starts AND the fight ends just when the last use ends. I for one don't like basing a gearing strat on something as variable as fight length.

Second, you are comparing the on-use proc to the static stat. Name two trinkets where that comparison is applicable. Most (actually maybe all?) +321 str trinkets have their own proc/on-use effects. Impatience of Youth will be a better trinket after the PTR not because of your calculations, but because its 321 mastery will be much more valuable.

Exemplar · 02/03/11, 9:56 AM

Chippy, very good and solid napkin math. I agree with both premise and results.

My only comments would be:
1) Proc trinkets should often pull ahead of static. Static X becomes 1.2X during AW, using perfect is 5/6X and 1/6(1.2X) = 1.03X averaged. Proc will be from worst case X (no procs overlapping with CD) to best case a full 1.2X. Since they are a large spike (rather than a static value) that large spike corresponding to any portion of AW will quickly pull ahead of the static.

If a proc item matches a static value when you average both disregarding CD, then the proc should actually be superior when CD are integrated.

So static/proc/use averaged values being equivalent the general sequence of desirability would be On Use, Proc, Static. I again stress, only if averaged values are equal.

2) Not that you even imply otherwise, but you never know precisely a fight's length until it's over. Assuming shorter than truth causes CD mismanagement, assuming longer than truth causes CD mismanagement. Obviously this is unavoidable. Sometimes you'll be spot on, sometimes use too early or late and fail to line up to best effectiveness or entirely lose a CD.

On average we probably end up with less CD usage than averaging math would suggest, but it's still the easiest/best assumption for all calculations.

Lorayth - True, there are two things to consider, Str and Rating. However, since Str is so far superior to other stats we should be focused more on static Str vs proc Str for this discussion. If Static Str > Proc Str, or vice versa, it's unlikely that the Static Rating vs Proc Rating in the same comparison would alter the bottom line assessment.

We cannot count on Pie in the Sky, which is why we use averaged values to determine comparisons. In the long term you get closer to the average than the best (or worst) case scenario.

Lorayth · 02/03/11, 10:23 AM

Originally Posted by Exemplar

Lorayth - True, there are two things to consider, Str and Rating. However, since Str is so far superior to other stats we should be focused more on static Str vs proc Str for this discussion. If Static Str > Proc Str, or vice versa, it's unlikely that the Static Rating vs Proc Rating in the same comparison would alter the bottom line assessment.

I agree with the premise of your statement Exemplar, but I know of no trinket (correct me if I'm wrong) where the static stat AND the proc are both str, which makes comparison of the static component to the proc/on-use component of two trinkets unrealistic. +321 str trinkets are vastly superior not because somehow that +321 str trumps over everything, but because the static stat is superior to the static stat of other trinkets that have a str proc/on-use (which in turn means they DON'T have a str static stat). For example, Hurricane is so much better than Impatience of Youth not because of any comparison between the +321 str vs. the str on-use, but because of the +321 str vs. the +321 mastery (which btw, as mastery stands right now, the on-use on Impatience of Youth would have to increase your str by something like 3.5k to make up for the difference). In the end, trinkets have to be evaluated with both components in mind, because as much as we like to think str is king, the other stats still matter (just not as much).

Hogun · 02/03/11, 11:07 AM

@Lorayth

I think the procs/on use/static trinkets have been exceptionally well balanced. I think we would all agree that a Str passive with Str secondary would be excellent but that is not the way they went with it. The trinkets have been balanced this time around so that you basically get the secondary as a flat stats ON AVERAGE. The reason that the Impatience of Youth is not that good after 4.06 is due to the trinket having an under budget str use not because of a passive that differs. It is more than ok to not have a passive STR stat on trinkets if there is a secondary of str. At this point it really comes down to preference. Given current balance of trinks I would prefer on use simply due to the reasons outlined by chippydip above.

When you say that trinket would need to have a use of 3.5k str I have to assume you mean under the old trinket system which just doesn't fit any longer.

Lorayth · 02/03/11, 11:19 AM

Originally Posted by Hogun

@Lorayth

I think the procs/on use/static trinkets have been exceptionally well balanced. I think we would all agree that a Str passive with Str secondary would be excellent but that is not the way they went with it. The trinkets have been balanced this time around so that you basically get the secondary as a flat stats ON AVERAGE. The reason that the Impatience of Youth is not that good after 4.06 is due to the trinket having an under budget str use not because of a passive that differs. It is more than ok to not have a passive STR stat on trinkets if there is a secondary of str. At this point it really comes down to preference. Given current balance of trinks I would prefer on use simply due to the reasons outlined by chippydip above.

When you say that trinket would need to have a use of 3.5k str I have to assume you mean under the old trinket system which just doesn't fit any longer.

Well I do admit that my 3.5k example is somewhat exaggerated because it's based on 1) the bugged Hurricane currently and 2) mastery being about as relevant as armor pen right now, and armor pen doesn't even exist anymore. However, that was exactly my point. The assumption everyone is making is that come 4.0.6, all secondary stats will pretty much become equal. History of this game and the ingenuity of the people on this forum have proven otherwise in the past. Eventually we will figure out what the "best" secondary stat is, and any comparison of trinkets would have to include WHAT the non-str component of either trinket are to be realistic. I'm not yet convinced that just a comparison of the str component of any two trinkets (be it static or proc or on-use) is the only thing that matters. Hurricane's proc, even after bug fix, is a great example.

In any case, this discussion is mostly an academic exercise as far as I'm concerned, so I will refrain from further nitpicking. Just want 4.0.6 to go live at this point

Edit: Perhaps Impatience of Youth being under budget might be intended instead of being a bug. Yes, right now it's worse than even a blue trinket, but I don't think that'll be the case once mastery is on par with other secondary stats. But it's also exceedingly easier to get than other epic trinkets. I personally don't have a problem with it being under-budget.

Exemplar · 02/03/11, 12:24 PM

Lorayth, while you could compare the static Str of Crushing Weight to the static Mastery of [Impatience of Youth] (variable Str vs Mastery), it is more useful to compare the static Str to the on use Str (variable of static vs on use).

Str is naturally ahead of Mastery, that's presently a no brainer. How much Str the Use provides is the question.

Weight is 321 Str. Average this over a fight with no AW: 321 Str.
Youth is 1605 Str on demand. Average this over a fight with no AW: 267.5 Str.
Variance: 53.5 Str, Winner Weight.

Now average them with AW:
Weight is 1 (active time) * 5/6 (time outside AW) * 1 (damage modifier) * 321 = 267.5
plus 1 (active time) * 1/6 (time inside AW) * 1.2 (modifier) * 321 = 64.2
Total = 331.7 Str

Youth is 1/6 (active time) * 1 (time inside AW) * 1.2 (modifier) * 1605 = 321.
plus zero (no time active outside AW).
Total: 321 Str

Variance: 10.7 Str, Winner Weight.

Youth made up a lot of ground based on linking with a single CD. GoaK could possibly cause Youth to fully tie or even pull ahead.

Perform the same with a Heart of Rage
Outside AW:
Heart average is 321.

Variance: 0, tie.

Inside AW:
Heart is 1/6 (active time) * 1 (time inside AW) * 1.2 (modifier) * 1926 = 385.2.
plus zero (no time active outside AW).
Total: 385.2 Str

Variance: 53.5, Winner Heart.

Next we should factor the values for Haste proc vs static Mastery or Expertise (with further juggling to account for reforging). Even if we state you get a ful average ofl 385.2 Haste by proccing only during AW (intentionally overvaluing), compared to 321 of another stat you only have variance of 64.2 Rating.
Variance 53.5 Str and 0 Rating for Heart vs 0 Str and 64.2 Rating for Weight. Winner, Heart.

Results: Whether an item is passive, proc, or on use does matter.

Lorayth · 02/03/11, 12:41 PM

Originally Posted by Exemplar

Results: Whether an item is passive, proc, or on use does matter.

The long post is appreciated but wasn't necessary. I never questioned that conclusion about on-use vs proc vs static.

However, your conclusion was not conclusive about trinkets until you added the second part. Now, compare Heart to Hurricane. The fact is the on-use str of Heart is better than the static str of Hurricane, yet we are all in agreement that the proc for Hurricane is so much superior than the haste proc of Weight or static expertise of Heart that it is better than both. Sure, you can say Hurricane is the exception to the rule, but given that we are all here to maximize our performance, which in big part means maximizing dps, doesn't the "exception" become the rule? That was really the reason why I responded to chippy's post in the first place.

Maybe I mis-read chippy's post. If it was a discussion about on-use vs. proc vs. static, then I apologize for my ranting, but when he used that word "BiS" purely based on that one calculation, alarms went off in my head.

chippydip · 02/03/11, 1:17 PM

Originally Posted by Lorayth

Two problems with your calculations. One, your calculations are based on pie-in-the-sky scenario of when the stars align and angels dance over your head, where the fight length is exactly that you can pop the on-use trinket when the fight starts AND the fight ends just when the last use ends. I for one don't like basing a gearing strat on something as variable as fight length.

Second, you are comparing the on-use proc to the static stat. Name two trinkets where that comparison is applicable. Most (actually maybe all?) +321 str trinkets have their own proc/on-use effects. Impatience of Youth will be a better trinket after the PTR not because of your calculations, but because its 321 mastery will be much more valuable.

Actually, most of the calculations are based on a fight length that's a multiple of 2 minutes so that if you popped cooldowns right at the start they would be almost-but-not-quite usable when the fight finishes. I mention that this 1/6th uptime for "Use" effects is actually pretty much the worst-case and then showed a best-case scenario just for funsies. The actual value of the trinket should fall somewhere in between.

As for your second point, its certainly possible to compare static Str to static secondary stat and proc/use secondary stat proc/use Str, I just think its easier to flip-flop and compare the static Str to the proc/use Str and then the proc/use secondary stat to the static stat. In either case, though, I'm not sure how you'd do the comparison without trying to figure out how much a "use" or proc effect would be worth as a static stat. The benefits I describe for GoAK are Str-specific, but the benefits from stacking with AW and Zealotry will effect any dps stat in pretty much the same way so you can use a similar analysis for crit/haste/mastery use effect if there were any.

The reason I even thought to run these numbers is people saying that Impatience is under budget when its pretty clear from looking at other "use" trinkets that this is the intended weighting for "use" trinkets. The budget for and ilvl 359 trinket is 2 stats of 321 item points each. For static stats its easy to see that things have the proper budget. Clearly blizzard thinks that a "use" of 1605 item points for 20 second every 2 minutes is "worth" 321 item points and that a chance of hit of 1926 item points for 20 seconds with a 100 second ICD is "worth" 321 item points as well.

I wanted to see how far below the ideal 321 budget Blizzard's weighting for a "use" trinket would put it, but it turned out that they seem to actually come out a bit over budget in typical fight conditions (and possibly way over budget if the stars align).

Originally Posted by Exemplar

1) Proc trinkets should often pull ahead of static. Static X becomes 1.2X during AW, using perfect is 5/6X and 1/6(1.2X) = 1.03X averaged. Proc will be from worst case X (no procs overlapping with CD) to best case a full 1.2X. Since they are a large spike (rather than a static value) that large spike corresponding to any portion of AW will quickly pull ahead of the static.

Yeah, this is something I didn't even pretend to have a definitive answer for. If I have time I might throw together a simple sim to try and isolate how a proc will overlap with cooldowns by chance and if you can do better by watching the proc.

Originally Posted by Exemplar

2) Not that you even imply otherwise, but you never know precisely a fight's length until it's over. Assuming shorter than truth causes CD mismanagement, assuming longer than truth causes CD mismanagement. Obviously this is unavoidable. Sometimes you'll be spot on, sometimes use too early or late and fail to line up to best effectiveness or entirely lose a CD.

Very true. Personally, I handle this by popping cooldons right at the start and chaining them from that point onward and then watching a Time-To-Death mod to try and figure out if I'll get an extra use by continuing to chain or if I'm only gonna get one more use and can wait a little longer to make sure I've refreshed Inq and/or anything else that might give me that little extra boost.

(At the start of a fight, I've been building to 3HP and then popping Inq and GoAK and then trying to judge when I'll have 3HP closest to 20 seconds remaining on GoAK to pop AW+Zeal and then use AW+Zeal on cooldown from then on and either hold my next GoAK until AW+Zeal have 10 second left on cooldown or until the boss has less than a minute or so to live. Using GoAK without AW+Zeal only becomes desirable if the fight lasts in the 10-11 minute range where you can get a 3rd use in, but only if you don't delay the 2nd use. Most fights don't last this long, so its generally not something I even worry about.)

Originally Posted by Lorayth

Maybe I mis-read chippy's post. If it was a discussion about on-use vs. proc vs. static, then I apologize for my ranting, but when he used that word "BiS" purely based on that one calculation, alarms went off in my head.

Sorry for the confusion. I was just trying to point out that "use" trinkets probably need to be re-evaluated. My understanding is that post-4.0.6 Hurricane is still BiS, but that its only marginally better than the various 372 trinkets. In a fight where Fury can be used with cooldowns efficiently it seems like its probably better than the 372 trinkets which means it may top Hurricane as well (and certainly any "use" trinkets in the next tier would best it). Until 4.0.6, though, Fury still sucks anyway so Hurricane is clearly still BiS.

Personally, I'm thinking about shooting for Fury + Hurricane with Impatience + Hurricane as a backup on fights where having to stack Fury for the use effect causes problems (Atramedes, Conclave, etc).

Exemplar · 02/03/11, 2:34 PM

Originally Posted by chippydip

Personally, I'm thinking about shooting for Fury + Hurricane with Impatience + Hurricane as a backup on fights where having to stack Fury for the use effect causes problems (Atramedes, Conclave, etc).

This is likely the best solution. Technically [Heart of Rage] should be better than Impatience in the second scenario, however the expertise on Heart (even reforged down) would mean a swap of just the trinket would put you drastically over Exp cap and be a significant loss of secondary stats.

Also I expect there will be less issue with Fury (post-patch with increased options to proc and longer duration on the stack buff) than expected. To use the Atramedes example you would lose any stack during flight phase. But if your next CD period is 15-20 seconds after he lands, odds are high that you could re-stack before its CD is completed. So there may be fewer situations needing a swap than at first glance.

Regarding DMC: Hurricane, its chance to proc damage does not compare directly or easily with either Str or Ratings. The closest you could get would be to assign it a DPS value and then translate any Str/Rating variance into pure DPS to compare side by side. That's more or less what I've done in my spreadsheet and what others have done with other simulation tools.

P.S. I'd err on the side of AW/Zeal earlier in GoAK than even a second late. Not-quite-fully-stacked Str bonus is better than extending beyond and no Str bonus at all.

Lorayth · 02/03/11, 2:46 PM

Originally Posted by chippydip

Personally, I'm thinking about shooting for Fury + Hurricane with Impatience + Hurricane as a backup on fights where having to stack Fury for the use effect causes problems (Atramedes, Conclave, etc).

License to Slay works out to an avg of 367 str, assuming a 5 min fight and 20 seconds to ramp up (380 * 280 + (0 + 380) / 2 * 20 ) / 300 = 367.3). This is on par with your avg calculations for Impatience even with AW + Zeal + GotAK, without the uncertainties that's built into on-use trinkets, AND this is without factoring in the benefits it gets from AW + Zeal + GotAK (which would be slightly better than the 6-9% buff to static trinkets, only because you'd pop AW + Zeal after it is fully stacked at 380 str; this works out to an avg str of 400+). The only fight I can think of where LtS would not be ideal would be Atramedes because of the air phase; even on Conclave you can make the jump during transition in less than the 15 seconds for LtS stacks to fall off. I'm rocking Hurricane and LtS myself, and am pretty happy with them.

Lorayth · 02/03/11, 2:55 PM

Originally Posted by Exemplar

P.S. I'd err on the side of AW/Zeal earlier in GoAK than even a second late. Not-quite-fully-stacked Str bonus is better than extending beyond and no Str bonus at all.

Usually I pop Inq and GotAK, then pop AW/Zeal once HP gets up to 3 again (while using up any HoL procs in the mean time with TV). Yeah, I'm probably a couple seconds from the most ideal use, but it's simpler than having to time it exactly.

Redcape · 02/03/11, 4:45 PM

I have an update on stat weights for 4.06. I caught a bug in my spreadsheet where I was calculating HoL damage based on CS/TV swing damage *without* crit and glyphs factored in. This caused both crit and mastery to clock in lower than they should have and also underrepresented our dps.

The current best estimate I have for stat weights is:

Weapon Speed 174522
Weapon DPS 872

Strength 232
Hit rating 175
Exp rating 134
Crit rating 94
Haste rating 67
Mastery rating 99

Of course, as usual, the Hit/Exp weights are for precap values, clearly they are really worth between 99 and 94 depending on how reforging math works out.

This is a substantial shift as it means that haste is noticeably worse than the other stats, mastery is definitely our best secondary stat and my current estimate for our dps improvement when patch day hits is 18.4k -> 21.4k for 16.3% improvement. I am also showing refreshing Inq with up to 4 seconds remaining at *any* amount of holy power as being optimal (though with the usual caveats that the difference is miniscule) and the optimal rotation as

Inq
TV
HoW
CS
Exo
Judge
HW
Cons

Again, the positioning of HoW and Exo is very weakly optimal, any combination of using Inq first, Judge/HW/Cons last is extremely close.

Note that all these estimates of changes to dps are ignoring the metagem and bracer enchant changes, which will add a little bit more.