Ayag wrote:Anyone has a source or know the maths behind the stats equivalent on the first page (hb/int/mp5) etc ?
Really curious about it, and while I don't say it's false, I find some values very odd, so would like to see the reasoning
Cheers
The numbers are based on the "standard" way of calculating mana vs healing and that is.
1 healing = (length_of_fight/average_time_between_heals)
×average_heal_coefficient
×average_uprank_mana_difference
/average_uprank_healing_difference
(Downrank values are normally used for maxrank spells due to lack of alternative methods)
1 crit% = no agreed upon standard method. I think he tries to convert it to +healing without taking either overhealing or armor buff into account. I personally just value it at 5 +healing (that's rather low, intentionally so based on the fact that my crits are often close to 100% overhealing) and move on

1 mp5 = length_of_fight/5
1 int = 15 + critvalue/59.5 (Times 1.1 if alliance with Blessing of Kings)
1 spi = c.b.a to type it out but how much mana is generated given estimated time in five second rule and talents, setbonuses etc + 0.25×the value of +healing. Then multiply by 1.1 if alliance (Blessing of Kings) and an additional 1.05 if human. (5% spirit racial.)
All values are then normally divided by the value for 1 +healing, to get HEP. (Healing Equivalent Points)
Obviously this method depends heavily on what assumptions you make for the different averages. The table quoted on page 1 assumes (if I remember correctly and I think I do) that the only spell used is Heal Rank 2 and that it is cast with 2.5s intervals for 2 or 15 minutes respectively with 0 time outside of the 5 second rule.
I find the values generated by those assumptions very weak and have instead made my own estimated with the data I've recorded from my own raiding. For me personally these values are obviously significantly more accurate, but I think that they are also more accurate for anyone who doesn't spam Heal Rank 2/3 without aborting.
I should probably update the numbers since the sample size used was really small, but here is a link to those values anyway.
Link here.