Originally Posted by Qieth
Pro haste:
More haste, even if we cannot get another starfire in during eclipse, would allow for faster starfires after eclipse, which would assist in proccing solar eclipse faster.
More haste means more starfire crits for our tier 10 4-piece bonus.
Pro crit:
More crit outside of eclipse allows for faster eclipse procs after both eclipse's.
More crit means more big starfire crits with our tier 10 4-piece bonus.
What is your take on this?
|
In this post I am assuming both soft caps as a starting point since both are readily reachable. I am also ignoring bloodlust, power infusion, and potions since they generally are harder to plan for. I realize that much of this may have been said before, but from what I've seen, not all in one place. I am also going to go out on a limb and disagree with the TTT statement that post-soft cap haste and post-soft cap crit have equal value.
We basically have 4 phases of our stationary rotation: SE, trying to proc LE, LE, trying to proc SE
SE: Given that NG lasts for 3 secs, we basically have 3 tries to proc NG again. With >55% crit NG will fail to be refreshed less than 10% of the time (0.45^3 = .09). So haste has very little value.
Trying to proc LE: Same as during SE, NG will have such a high up time that additional haste has little value.
LE: The only real value haste has here is to squeeze in an extra buffed SF. Here's the formula I've been using:
Haste Rating Req = {3 / [1.2 * 1.05 * 1.03 * 1.03 * (15 - TL) / X] - 1} * 100 * 32.79,
where TL is the time lost on an extra wrath cast (i.e. poor play), latency, and/or instants (dot refreshes, starfall, on-use trinkets, pots, barkskin, innervate, etc.), and X is the number of starfires you want to squeeze in. For me personally, I typically lose between 2 and 2.5 seconds (an extra wrath cast because I'm bad, a DoT refresh and about 150-250 ms of lat). Given this range of time lost, I would need to acquire well over 700 haste to fit in an additional buffed SF. So for me at least, I gain nothing from additional haste in this phase. Skilled players that never waste that extra second will lose between 1 and 1.5 secs if they have to cast an instant and have some degree of latency. These players can squeeze in an additional buffed SF nuke by hitting 510-540 haste.
There is an additional complication that might have arisen from the way Blizzard has broken the WiseEclipse mechanic (or at least from what I understand of how they broke it). If you over shoot the amount of haste you have and your last SF cast that was started with eclipse but ended without it lands in the "unproccable dead zone", you will have a wasted SF cast that wasn't buffed by eclipse and had no chance to proc the next one. At least it has happened to me where I crit with a SF cast started under the effect of eclipse but had buff ran out before it ended and still failed to proc the next eclipse (I'd like to hear some outside confirmation of this happening though).
Trying to proc SE: This phase can be more or less be modeled as a geometric distribution (there is a minor complication in that only the first 2 SF casts will be under the effect of NG, but I don't think that this affects the result). Napkin math inc...
The expected time until your next SF crit = (1/CP) * CT, where CP is the crit % and CT is the cast time. Plugging in the numbers:
Exp. Time = 1 / [(1.85 + INT/166.67 + CR/45.91)/100] * 3 / [1.2 * 1.05 * 1.03 * 1.03 * (1 + HR / 3279)]
simplifying and assuming a very generous raid buffed INT of 1900...
ET = k / [(85 + 0.275 * INT + CR)(3279 + HR)], where k is some constant.
finding the marginal values of CR and HR to reduce ET...
∂(ET)/∂CR = - k / [(607 + CR)^2 (3279 + HR)]
∂(ET)/∂HR = - k / [(607 + CR) * (3279 + HR)^2]
For any value of CR (700-1300) and HR (400-1000) that are attainable, (607 + CR) < (3279 + HR), which means that |∂(ET)/∂CR| > |∂(ET)/∂HR|.
In other words, point for point, on average crit rating will do more to reduce the waiting time before the next eclipse for all values of crit and haste you could possibly have.
TL;DNR: Haste only really comes into play during LE and when trying to proc SE. During LE, you want just enough so that for your rotation/play style you can maximize the number of buffed SFs you cast, but you don't want to over shoot as this can result in unbuffed SFs that can't proc the next eclipse. Although haste does help you proc SE faster, point for point crit will help you proc SE faster more than haste. Basically, I plan on shooting for some ideal range of haste values that fits with whatever I generally do during LE, and then go for crit the rest of the way.