5.2 Trinkets Discussion

[whathump]

And every target switch resets your swing timer, which also decreases the value of haste, as it decreases your autoattack and thus poison damage.

I don't know the hard maths behind this reasoning, but feel free to browse Method's rogues for example. Mastery>Haste>Crit seems to be the way to go.

I also guess that going with mastery will decrease maximum dps but increase average dps as its less rng, while a haste build relies on the unreliable rppm stuff.

Good rogues always go for safer and reliable damage rather than potential higher dps.

I figure it's good to keep mastery and haste about equal when you have the meta gem because of how powerful it is.

EDIT

Looking at my logs, I see that my average Juju uptime has been consistent with what it should be for my haste level (~20%), with sometimes being higher (30-40%) if I get good luck, but almost never lower. I don't usually get the chance to wait the 5 minutes to guarantee a proc in my guild.

[theherecy]

Good rogues always go for safer and reliable damage rather than potential higher dps.

We're quite possibly the most stable class there is! I would like to get some maths on it, mastery spanks haste on aoeing (bats on tortos, etc), but when it comes down to tab rupturing I'm just not 100% sure.

Also to note, it's not like the mastery > haste > crit is something new. But with method's rogue the more and more mastery there is and the more he "specs away" from haste, the quicker his dps will fall. Essentially he will become like the rune of reorigination. I don't necessarily think having a haste/mastery blend is inconsistent, and if that were the case fire mages just wouldn't make the cut.

Breezing over my logs (haste/mastery) everything is fairly consistent between kills and wipes.

[Deleted]

Good rogues always go for safer and reliable damage rather than potential higher dps.

?
Define good rogues.

A quick word about Rune: I quite like it. Granted, I don't think it's as good as Renataki's or Talisman, and its design is indeed somewhat clunky. But it has the merit to be really innovative, and I feel it is the trinket which minmaxing its procs will have the biggest impact, among the five of this tier. It is however very unstable from one pull to another, much more than the standard RPPM trinket (which already is way more than any trinket with classical design), and syncs very, very badly with Talisman.

[fierydemise]

Bear in mind that Shadowcraft models a great many number of things, but its trinket model isn't quite as refined as the one on this thread. For example, it doesn't account for trinket value variations with fight lengths (and so nor does it account for the increase in EP value if waiting 5 mins before a pull).

The biggest issue with the shadowcraft model live currently is that it doesn't take into account the effects of bad luck protection for RPPM trinkets and has a few other things not ideally handled. Changes to improve the RPPM modeling substantially are in the 5.3 experimental build but that isn't on the website yet and won't be probably until 5.3 comes out.

?A quick word about Rune: I quite like it. Granted, I don't think it's as good as Renataki's or Talisman, and its design is indeed somewhat clunky. But it has the merit to be really innovative, and I feel it is the trinket which minmaxing its procs will have the biggest impact, among the five of this tier. It is however very unstable from one pull to another, much more than the standard RPPM trinket (which already is way more than any trinket with classical design), and syncs very, very badly with Talisman.

Perhaps I'm not seeing but I don't see what rogues have that would work well to minmax Rune procs. The two specs that get a lot out of Rune are specs with big snapshoting mechanics that rogues have none of. Additionally the increased value of haste especially with the legendary metagem potentially complicates rune as well since I suspect gaming Rune procs involves a mastery proc which would otherwise limit the degree which you could stack haste.

[wimp]

Regarding rune: I'm afraid I share Fierydemise's lack of vision here... The only 'gaming' I've come up with for RoRo is to use it in synergy with Talisman with a build that balances mastery, haste and crit, with haste just ahead of the rest.

With some very rough assumptions, a RoRo proc gives you 50% haste for 10 seconds, which significantly increases Talisman's value (as well as the dps from meta-gem procs). But I don't think this approach will be quite good enough: Talisman won't always magically stack to 5 in those 10 seconds.

Note that while RPPM trinkets scale roughly linearly with haste, the scaling is dependent on the proc rate (for the visual mathy types: the gradient of the uptime vs haste line increases with the RPPM value): those with a high proc rate, like talisman, scale very well with haste,while those with a rubbish proc rate, like RoRo, don't really benefit a huge amount from haste. What this means for us is that RoRo will make Talisman quite a bit better. But Talisman doesn't really change RoRo much (and since RoRo has an ICD, RoRo doesn't buff itself in the way Talisman does).

With a heap of assumptions mixed with napkin math, I make Talisman be worth around 500 EP more if used with RoRo in the way described above. Which makes me think is not worth dropping Renataki or Bad Juju for RoRo. Though it won't stop me playing around with it once I do get me a Talisman, in case there's something I've overlooked/any assumptions which don't hold.

I can't think of any RoRo mastery proc uses which would make RoRo good. Having no haste is no fun at all... But do please say if you've found something clever to do with RoRo.

[Ryme]

?
Define good rogues.

A quick word about Rune: I quite like it. Granted, I don't think it's as good as Renataki's or Talisman, and its design is indeed somewhat clunky. But it has the merit to be really innovative, and I feel it is the trinket which minmaxing its procs will have the biggest impact, among the five of this tier. It is however very unstable from one pull to another, much more than the standard RPPM trinket (which already is way more than any trinket with classical design), and syncs very, very badly with Talisman.

Rune is a trinket that I would actually like to see either off the RPPM system and back on the old ICD system, or an on use effect. Obviously, things would have to be scaled appropriately to compensate, but if you could accurately predict this trinket's behaviour fight to fight, it would hold some potential - in the least, it would be more compelling to math out.

---------- Post added 2013-04-26 at 04:15 PM ----------

My maths still has Talisman coming out far too low, to check I'm getting the same results as you wimp, would you mind verifying the following for me on a 541 Talisman with 16% haste?

EDIT: No need, it's horrible in so many ways, it needs a complete rewrite - This trinket is becoming the bane of my life, for the life of me it will not behave - I'm getting total % exceeding 1, uptimes looking like they're based upon sin functions, it's pissing me off no end.

To give me some comparison, how are you physically modelling the trinket? Are you taking % uptimes for each stack as I was? You seemed to imply you weren't but I never got any indication of the route you went down, that or I missed it.

[Deleted]

?
Define good rogues.

well, the ones with a safe spot in all the top guilds over the years. Usually they for example cap expertise even when ignoring it would gain 1-2 % potential dps.

[wimp]

@ Ryme: I think you've done all the hard work already in calculating stack uptimes - once you're there, there's not much more I've done. For the 541 Talisman at 16% haste I get for the stacks 1 to 5:

27.52%
13.01%
6.30%
3.13%
3.38%

(and from our exchanges a couple of pages back, I think we're on the same page up to there).

Once you have uptimes for each stack, you just multiply percentages with haste values and add them up. So (% stack 1) * (haste stack 1) + (% stack 2) * (haste stack 2) etc. For the 541 talisman, this gives a total of 3417 haste.

You could leave it at that (i.e. 541 talisman gives 3417 haste with 100% uptime, move along, nothing else to see here). But I do one more thing, to force the talisman to fit into the same spreadsheet as all the other ones: 3417 haste with 100% uptime is the same as 6407 haste with the talisman's uptime of 53.33%. So that figure of 6407 is the one that goes into the spreadsheet.

That's assuming an infinitely long fight.

(I've then tweaked that figure assuming a guaranteed proc on pull, which results in adding an uptime amount dependent on the overal fight length. For the talisman's case that's a very small difference - it's only really significant for Renataki's and Bad Juju).

[Ryme]

My results are slightly off:



Anything immediately strike you as... off?

[wimp]

Could it be that you're taking the extra haste given by the first stack to increase the proc chance of the first stack? I have the haste given by the first stack increasing the proc chance of the second stack. I *think* that could account for our differences.

[Ryme]

No, that seems fine. Though, I seem to be making no use of lambda5, which worries me since that should be my "refresh" lambda.

---------- Post added 2013-04-29 at 10:37 AM ----------

So, I've redone my sheet to account for a 5 stack refresh, my new numbers:

1 - 0.2750
2 - 0.1300
3 - 0.0630
4 - 0.0312
5 - 0.0517

Which does seem to fall closer in line with the simcraft results:

1 - 0.2829
2 - 0.1500
3 - 0.0819
4 - 0.0459
5 - 0.0712

[wimp]

Moving the spreadsheet differences geek chat to PMs. As much as I love excel, I'm often reminded much of the world's population is ignorant to its charms...

Ryme: how are you making simcraft show you uptimes for the different stacks? I can only see the overall uptime. Also, do you/does anyone know if simcraft has RPPM trinkets proccing on the pull (I would have assumed so, but want to check).

[Nitwit]

Moving the spreadsheet differences geek chat to PMs. As much as I love excel, I'm often reminded much of the world's population is ignorant to its charms...

Ryme: how are you making simcraft show you uptimes for the different stacks? I can only see the overall uptime. Also, do you/does anyone know if simcraft has RPPM trinkets proccing on the pull (I would have assumed so, but want to check).

Under the buffs section, click the little arrow to the left of a buff's name for expanded details (for talisman, this will include each individual stack uptime).

From memory, SimC assumes that you build up 5 minutes of bad luck protection before the pull (might be misremembering, though).

[wimp]

Under the buffs section, click the little arrow to the left of a buff's name for expanded details (for talisman, this will include each individual stack uptime).

Wah, clickable... thank you very much! All this statistical data is pretty hot stuff...

From memory, SimC assumes that you build up 5 minutes of bad luck protection before the pull (might be misremembering, though)

Thanks. Would be good to get a definitive answer (if someone knows), but will take this to be true for now (and I'll try to look at the code at some point).

***

I've been playing a bit more with simulationcraft, and not entirely surprisingly, I have a question. Simulationcraft has the 5% spell haste buff affecting our RPPM mechanics: is there something I'm missing about spell haste affecting rogue RPPM mechanics? I think this is a bug, but I thought I'd check in case I'm being thick.

[Pathal]

Our spell haste is higher than our melee haste since it only works off true haste now.

[wimp]

But I thought the 5% spell haste buff wasn't considered "true haste" either? Or am I wrong here (if so I'd be grateful for a correction!)

I've been taking Haste = haste from gear (+ haste from bloodlust, the overall figure of haste % this adds being dependent on fight duration)

Generally confused trying to hastily sort hastes into true and lying sonuvabitch hastes...

---------- Post added 2013-04-30 at 04:34 PM ----------

Update. From the Simulationcraft people: "(...)we verified this with Blizzard and the spell haste buff is for some reason true haste(...)"

[Pathal]

Yeah. They have a horrible track record with differentiating the two that I'm never sure anymore.

While I'm here, remember that Berserking (troll racial) is true haste as well.

[Tehstool]

Basically, the rule of thumb is that anything that says haste is true haste and anything that says attack speed isn't.

[RemJay24]

So the Rune isn't great for ALL Rogues or just mut/combat (which I agree, is nearly 99.9% of all Rogues)?

I'm a fan of Sub PvE, though for progression sake - haven't played Sub pve for a while. Does the trinket pose any benefit more so for Sub than the other two specs? Or is the trinket a dead duckling for all three specs? Hard to judge off Shadowcraft, as Sub isn't as nearly accurate on there as combat/mut, but for good reason.

Thanks!

[theherecy]

So the Rune isn't great for ALL Rogues or just mut/combat (which I agree, is nearly 99.9% of all Rogues)?

I think it'll be better for other classes. Poisons rely on a good combination of haste and mastery and assassination depends on poisons. The other specs don't care too much for them, so it may make the trinket better. Combat would be a weird one, you want haste always and the trinket will either give you so much you will cap easily or take it away