It's not just the variance in start time, it's the minor variance in delay between each cast. It's small due to spell queueing, but nonzero.

So you if you want to be really pedantic about this, you'd have to take the the distribution in starting delay for the first Starfire, and then convolve it with the delay distribution 7 times (or invoke the Central Limit Theorem). Regardless, I don't see any reason why the mean continuous DPS isn't a perfectly good estimate.

I'd say it is! The only problem here is that the sim has symmetric and normal timing distribution curves for all of the timing values (and I'm not sure it uses any variance in travel-time at all), which causes the simulated value of haste to have a larger step function component than the real value should, which in turn pushes people to value marginal haste improperly. It's the kind of artifact you get in a lot of physics simulations as well, people just have to learn to work around it.

I suggest that people trust the spreadsheet for their marginal haste values instead.

Apologies for the improper use of discretely, it's a very bad habit I've picked up from talking too much about modelling with non-mathematicians. I was thinking along the lines of the messy approach, draw initial delays from a few different kinds of distributions and seeing if there was an obvious favouring between one type of distribution and real datasets but you're quite correct it'd be entirely useless as it would automatically fit the data you had and not necessarily give any insight into the problem.

You're entirely correct, I think. Too much liking of an idea before I've thought through the obstacles on my part I'm afraid.

Actually, just to expand why I was so taken with the idea: I geared up fairly quickly and saw my dps change strangely as I got from far under to hitting the haste soft cap. One raid I was reliably doing X, I got 100 haste ish increase in gear and the next it my dps had ratcheted up by far more than I'd have expected given continuous modelling of dps (it was going from 3.5 to 4.2K on average or something like that). It was strange and stuck with me and I attributed it intuitively to some haste threshold being passed for me on my connection etc for total casts within an eclipse or similar. I never properly discounted other factors though at the time so it could just be "noise".

On the subject of haste, in 3.3 [Formula: Enchant Weapon - Black Magic] is being changed to a haste proc (250 haste for 10 seconds, on a 35-second internal cooldown). The theorycrafting for other nuking classes indicates that this may be an improvement on 63 spellpower or even 81 spellpower for a staff. Would it be an upgrade for moonkins as well? Given how much time we spend in Solar Eclipse, it seems like it would be less beneficial for us.

Edit: Supposedly the ICD is a maximum of 35 seconds. It's being modeled elsewhere at approximately 30 percent uptime. Wouldn't that be closer to around 83 average haste rating?

It's won't be an upgrade. Since the proc is random, we can average it out as 50 haste rating - it'll be distributed over wrath casts and starfire casts in the same ratio that regular haste rating is (almost. Since the first proc will most likely be for a lunar phase, it'll have slightly higher average value for shorter fights). 50 haste rating is nowhere near 63 spellpower at modern gear levels.

In principle it could be on a second weapon swapped in just before a short burn phase (if your Eclipse status is right), or a Hurricane phase.

A 10 second proc makes the timing on that extremely tricky. Swap too early, and you may waste half the proc. Swap too late and rng may waste the entire proc.

Item - Icecrown 25 Normal Caster Trinket 2 - Each time one of your spells deals periodic damage, you have a chance to gain 918 spell power for 10 sec.

If there's a low ICD on this trinket, hopefully I will be able to pick this one up quickly.

Will almost certainly have the standard 45s ICD.

Edited.

Wow 10 man heroic trink same as illustration, that's kinda... Horrible?

It's not horrible, the trinket is still really good if not BiS, and not everyone has access to Illustration. Think of it as a levelling out of the playing field.

When is it time to move out of the t8.5 4p?

Right now I'm sitting on [Helm of the High Mesa] and [Robes of the Shattered Fellowship]
and I also have a Trophy of the Crusade to buy something.

Is it worth using these and letting go of the 4p?

I don't think I'd have an easy time convincing 9 other people to go farm it with me just so *I* can get a trinket.
It's time when you have 2T9, go with 2T8/2T9. If you don't expect to get a second trophy soon, buy a 232 piece with badges.

I could do it very easily, except 5+ of those who would come would also be after the trinket. It's a 20% drop rate, so every 5 weeks or so one of us could get one. Then convincing that person to stay doing the raid for the other 4 to get one would be difficult at best.

I remember that for the Coliseum they only listed the "proc" effect on trinkets like Death's Choice / Verdict, and some raw stats were added later.

Considering the Ilvl gap between [Illustration of the Dragon Soul] and the new trinket, we will probably see a "+80~ crit/haste/hit" in addition of the stacking spell power.

I just hope it will be crit

"4 pieces - Your critical strikes from Starfire and Wrath cause the target languish for an additional 5% of your spell's damage over 4 seconds."

I am curious how this will play out, wouldn't it be getting overwritten regularly thus negating the full value of the dps gain?

EDIT: Sorry I see this was answered a couple pages back. Please ignore.

Inbetween Hurricanes is a bad time to DoT, you can waste Omen of Clarity procs that are much more vital to maintaining Hurricane spam. I basically reserve every cast after a Hurricane to be either Starfall or Hurricane, until adds die anyways.

I also found that the Starfall glyph isn't really much better than IS glyph. You have to occasionally force yourself to single target Anub, either for mana reasons or JoL reasons or in that burn part of p3 where you switch off adds to kill Anub.

I've had decent luck on heroic Anub with stacking trinkets. You have enough time before the first set of adds to get full stacks, or close to it. Whether you do after burrow phases depends on the individual details of how your guild moves and kills scarabs.

Anub doesn't really hit that hard. It's nice to have IS up for the debuff, but I wouldn't take AOE breaks to do it. I gave a higher priority to keeping up the debuff in P3.

I've been raiding as moonkin for sometime and wondering about the best rotation for togc10? Any ideas, mainly on last boss.

Anub depends a lot on your raid makeup and whether you have to shoot down Orbs. If you have 2 or more good Warrior/DK/Rogue, the adds will die fast enough you're probably best served by Moonfiring each of them once and then continuing the regular single-target rotation on Anub, adding in Starfall and Typhoon (if specced). They're just not a big deal. If you do have to DPS the adds, I would lean towards single-targeting them while keeping DoTs on Anub, but I have no math or anything beyond a gut reaction to stand that on.

I am surprised by the mention of glyphed Typhoon for the ToGCs.

I am currently raiding in ToC10/25 and working through ToGC10 up to the twins. The initial impressions I had of typhoon was that typhoon is not a dps boost but more situational talent.

Since my group has better control over the factions fight, I removed typhoon from my talents but as I am understanding in ToGC10 it is worth talenting in for with the glyph? Or only if you have the points to spare?

It's pretty much only for Anub, as you have 5 targets and you want to keep NG up for Hurricane.

It's also semi-nice when moving and when you have dots rolling, that's pretty much it.