Elitist Jerks Resto gear weightings spreadsheet

 07/07/09, 4:24 PM #2 Erdluf Great Tiger   Erdluf Night Elf Druid   Echo Isles I think the overall approach is reasonable, but there are several issues with the details. You use per-cast SP coefficients. You use per-fight amount healed. Suppose my average non-crit Regrowth heals for 15k, Rejuv heals for 12k, and Lifebloom (non-stacked, non-crit) heals for 8k. Also assume that effective healing is proportional to total healing. Supposing my fight had total effective healing percentages of 50% Rj, 30% Rg, and 20% Lb. My number of casts would be proportional to Rj 50/12 = 4.2 Rg 30/15 = 2.0 Lb 20/8 = 2.5 Total = 8.7 Percentage of casts: Rj = 4.2/8.7 = 48% Rg = 2.0/8.7 = 23% Lb = 2.5/8.7 = 29% and the average per-cast coefficent should probably use these weights, instead of the original weights. Because of the way you use your average coefficient, it should argualbly be per cast-time, rather than per cast. Other items: 1) Norfair's "fully talented" coefficients don't include Nature's Splendor (divide his total coefficients by his per-tick coefficients to confirm). I think Splendor is a safe bet for most trees. 2) The wild-growth coefficient is a per-target number. It should be multiplied by six. 3) Your calculations on splitting the coefficients for Rg and Lb aren't "fair." You could treat Rj as six smaller spells, each doing 1/6th as much damage and having a coefficient 1/6th as large. Doing so would substantially reduce the average coefficient, but the reduction would be meaningless. 4) Your computations for the marginal value of haste and crit (and the int contribution to crit) assume that spells have no base damage. If I set SP to zero, the values of all three stats goes to zero. If you fix this, you might want to go to base damage per cast-time, rather than base damage per cast. 5) You treat Lb as if it contributes to Nature's Grace and Living Seed. 6) The MP5 model is too simplistic. If I change the fight duration, the values of SP and MP5 don't change at all. Changing the fight length from six minutes to one minute drops the SP contribution to healing by 1/6, and drops the extra mana generated by MP5 by 1/6. To be meaningful, the "Healing per point of SP", needs to stop assuming spam-casting for long fights, and the "Healing per point of MP5" needs to realize that you are GC bound for short fights, and for those fights extra mana contributes nothing. 7) It looks like your spirit-regen formulas are using the old 30% Intensity. It is now 50%. Good luck.
07/21/09, 5:21 AM   #3
Orlok3105
Glass Joe

Tauren Druid

Hellfire (EU)
 Originally Posted by Erdluf I think the overall approach is reasonable, but there are several issues with the details.
First of all, thanks a lot for feedback. I've tried to fix most issues you've mentioned and uploaded updated version of spreadsheet (link in first post is updated).

Major concern:
I've added per-fight amount healed conversion to per-fight number of casts, as per your example.

Other items:
1) Fixed.

2) Fixed.

3) I'm not (totally) understand your point here. I've added GCD/cast time/effect time values into calculation to get "fair" influence of haste, but when you cast, for example, Rejuvenation, you spend mana and cast time once, not 6 times over 18 seconds. Please explain what you meant by "unfair" splitting of coefficients.
*EDIT* I've introduced HoT clipping and overhealing support, that should make SP coefficients more fair overall.

4) Fixed.

5) Fixed.

6) I could use a hand here. I know that SP/mp5 balancing is (very/quite/a bit) flawed, but that's the best idea I have at this time, please spare your thoughts about how to fix this.
*EDIT* I gathered some data from druid healers in my guild and other raiding guilds about relation between length of fight and how long druid is spamming heal every GCD. Though evidence is a bit anecdotal because of small amount of samples, it is still showing that between 3 and 8 minutes % of presence is some kind of parabolic function.
There is a formula I'm using to determine portion of fight druid is not spamming heals:
Idle(t) = 0.0028*t^2 - 0.0105*t + 0.0059,
where t is length of fight in minutes.
So I've decided to use [Idle(t)*10] as a multiplier for MP5 contribution. I'm not sure if it is best way to solve this problem, but it is better than nothing.

7) Fixed.

Last edited by Orlok3105 : 07/24/09 at 5:39 AM.

 Elitist Jerks Resto gear weightings spreadsheet