WrathCalcs models the RPPM trinkets pretty well with the uptime distribution Hamlet put in for it - in fact, I wouldn't be surprised SimCraft uses the same formulas to check for when the RPPM trinkets/meta proc per unit time during their calculations (side note: Hamlet used a poisson distribution to account for procs overlapping, more info here: http://iam.yellingontheinternet.com/...advanced-rppm/). We have to remember that WrathCalcs is a steady state of our DPS over a set amount of time, without taking into account movement/stacking of certain cooldowns. It still stacks Inc/CA/NV, so most of the stacking is apparent. But you'd be correct in saying trinket procs get averaged instead of applied directly into cooldowns. I understand a tool like SimCraft can do that, but then again we're assuming SimCraft has the correct priority for boomkins and stays up-to-date with theorycrafting changes we are constantly revising, including the recent opener change.
What I'm trying to get at is both tools are relevant to our needs. WrathCalcs is good for quick gear changes, modelling our steady state DPS for any baseline fight, and is good to springboard logical ideas from, but lacks in the area of stacking buffs outside of our own as well as not showing hard numbers for multiple targets. SimCraft is great for looking at specific fights with movement and additional targets, can model the multiplying effect of buffs overlapping, but isn't always up-to-date with our spec and might have other shortcomings I'm not aware of.
What's funny is that SimCraft generally agrees with the results of WrathCalcs when you extrapolate the ideas into the dimension of more targets + light movement. In average normal gear (~523 ilvl), if we have a look at the marginal stat changes as we increase targets we can get a better feel for the symbiotic relationship between the two models. I've also included total DPS and SS proc waste for fun.
One thing to note is that I needed to change buff.shooting_stars.react to buff.shooting_stars.up in order to get reasonable amounts of SS proc waste. On top of that, I also added "interrupt_if=buff.shooting_stars.up" to the end of the four Starfire/Wrath lines, so that it properly interrupts our hardcasts if we get a SS proc. Without these changes, SimCraft reported 88% SS proc waste for 4 targets, which is preposterous. This largely devalues the value of crit, so the default SimCraft isn't as accurate as you'd believe. SS proc waste with the changes is still pretty high, but reasonable.
Crit build (5802 haste, 11894 crit):
targets 1 2 3 4 DPS 154648 204406 230201 255011 Int 5.48 7.18 7.99 8.83 Haste 3.51 5.14 6.15 7.22 Crit 2.51 3.43 3.48 4.10 Mastery 2.50 3.02 3.46 4.32 SS proc waste 24% 37% 56% 67%
Haste build (10307 haste, 7389 crit):
targets 1 2 3 4 DPS 159190 208607 236659 257616 Int 5.57 7.39 8.22 8.93 Haste 2.65 3.00 2.84 2.44 Crit 2.72 3.77 3.98 4.15 Mastery 2.53 3.35 3.55 4.14 SS proc waste 15% 32% 48% 60%
So after an hour of simming, what can we conclude from these tables? For a crit build, we can see the value of haste is much higher than the value of crit per point, and this gap is widened by the number of targets. This is a very significant outcome, especially since haste even approaches the value of int per point. SS proc waste is high with a crit build, as you would suspect, but this may be due to some unknown inadequacies in SimCraft. Regardless, they are within the realm of reason.
For the haste build, SS proc waste is almost exactly what I've seen on World of Logs for boomkins running a haste build. If we say the average amount of targets on Council is 3, then some anecdotal evidence from my last Council kill is spot-on (48% recorded SS proc waste - http://www.worldoflogs.com/reports/r...?s=4891&e=5154). Additionally, we see haste decreasing in value per point as targets increase, which is exactly what we expected to happen. Since we have already reached the GCD cap for haste at this point, more points of haste don't help to speed up our rotation if we spend most of our time using SS procs or DoTing. Additional haste is great when you aren't at the GCD cap since it allows you to DoT faster and hopefully reduce SS proc waste with the shortening of the GCD, but once you hit the 10296 cap more points wont contribute to that factor at all. Therefore, after the cap, haste is helping us mostly on our hard casts, but we're getting so many SS procs with more targets this too is reduced. Thus, we expect the value of haste to diminish as more targets are added when already GCD capped. We also see crit being our best stat with a haste build as well, with mastery close behind.
After looking at this data, we can see that SimCraft has essentially agreed with all the extrapolation we made from WrathCalcs. We have been looking at the base single-target, no movement modelling in WrathCalcs and finding haste is significantly ahead of crit until the GCD cap, whereby crit moves ahead. We have also taken crit to increase more than the other stats when considering more targets, which is not entirely supported by SimCraft, more so our own logical conclusions. Regardless, once you are capped, crit is the best stat you can stack, though I'm not sure on how accurate SimCraft is being with mastery. It's understandable that SimCraft shows mastery to be higher than purported in WrathCalcs since it would be benefitting from the simulation's stacking buffs and whatnot, and it reasons that with more targets you spend more time eclipsed (SS procs power through non-eclipse phases). Might need to ensure the modelling SimCraft does is completely accurate before we draw any conclusions on that point.
Anyway, lengthy post and mostly directed generally. I think the biggest thing to take from all this is not the actual DPS numbers between the two builds, but the vast differences between the marginal changes in stats. Int seems disproportionately large in the haste build, but that may be true. Haste is quite a strong stat, and does indeed scale well with more targets.
As always, let me know if you see any gross errors and I'll work to correct them.