What I did was 295 * (1 crit percent / 100) = 2.95 rating. I guess if you multiply your non-haste gear crit percent by 2.95 you'd get your pseudo-haste rating. As long as your haste gear has more rating than that, you should wear haste gear instead. I'm not sure if these assumptions are well-founded. The only thing I don't take into account is bonus crit damage.

Responding to the Crit vs. Haste discussion a couple posts up:

First, NG is worth 314 Haste, not 295. NG reduces Starfire's cast time by 20%. 20*15.7 = 314. Luckily, since NG is factored before any haste effects are applied, it can be reduced to simply haste. I.E. No matter how much haste you have, NG is always worth 314 more on the casts it affects.

Second, your rough calculations aren't quite accurate. Both Haste and Crit have diminishing returns, I.E. 1% +crit is worth 1% DPS iff you start out with zero crit chance. If you start out with 99% crit, +1% crit is only worth .503% DPS. Similarly with haste.

Here's a more accurate way of comparing Haste and Crit:

Assume
,
with being your expected (average) damage, and being your average cast time.

The percentage benefit of 1 Rating on DPS is given by

where is your current rating of either Haste or Crit, whichever you're trying to find. The point where they're equal is then when


We'll find the benefit of Crit first, since it has two parts: Damage and speed.




is the base average damage of the spell, which is 647 for Starfire after adding in Moonfury, is the damage coefficient, 1.2 for Starfire with Wrath of Cenarius, is your +damage from gear, is the Crit rating coefficient, 1%/22.06 = .0004533, and is your Crit rating (divide your total Crit% by 22.06 to get your effective Crit rating).

The affect of Crit on Expected damage is then

which, as expected, depends on your current Crit rating.


where is the Haste Rating conversion 1%/15.7 = .0006369, and is your haste. Note that this does not include the 314 effective haste from Nature's Grace, that's taken care of in the top of the fraction.

The effect of Crit on average cast time is then

which again implies that the more Crit you have, the less percentage benefit you'll get from adding more.

Finally, this gives us that


Now for Haste:


We already have the formula for time in terms of haste, so


Since it's more convenient, we'll break this down after reciprocating it, since that's what we're doing for the DPS calculation anyway.

which looks strikingly like the affect of Crit on expected damage. Which makes sense because they perform the same function.

For simplicity, we'll replace all instances of with , since you're more likely to know what your overall Crit percentage is than what that equates to in rating. Therefore, the affect of Haste on DPS equals the affect of Crit on DPS when


Since both Haste and Crit have diminishing returns, you get more benefit from haste when your current haste is LESS than the formula. I.E. if you have high crit and low haste, it's better to add haste. For example, say you have 25% crit on your character sheet, +4% for Focused Starlight on Starfire. You would need less than 751 Haste rating for 1 haste to provide more benefit than 1 Crit. If you increase it to 33% character sheet and add 4pcT6, for 42% Crit total on Starfire, you need less than 926 Haste rating.

i am not sure if i follow what you did here, the top of the fraction calculate the average cast time of starfire base on your crit. Why would you need to divide that by the 1+Kh*H

So, wait: Is there any chance we're close to having a closed-form equation for determining the relative values of the major stats (dmg, crit, haste, and hit)? Or are these just sort of over/under "haste is better until you reach XXX rating at Y.YY% crit"? Sorry, but (ironically enough) equations make my brain hurt. Doubly so when they're in math/physics terminology & fonts vs. programming/spreadsheet speak.

I mean the macro on the spreadsheet is certainly one way of approximating (and seems to be about the best way short of a simulation for determining regen effects) the relative values, but closed form equations would be quicker and easier to update, and would come in handy for a sort of gear-suggestion function I've been considering adding.

Also, the best way to use Moonfury is to apply it only to the spell damage as a coefficient--base damage should come from a look-up table based on the number of talent points spent in Moonfury, as using the talents as a multiplier doesn't accurately predict the in-game base talented damage.