Author’s Note: This post is a bit disjointed, as first I go into my methodology for stat weight estimation, and then into the state of Warlords. The first section of the post is on the former, and is a formulation I have used in the past and will continue to use on my spreadsheets throughout the expansion. I would LOVE suggestions and criticism to the math. The second part of the post is my more opinion-based reaction to the current state of Warlords stats, and I would like to see readers’ thoughts on this as well, but please separate reactions to that from reactions to my mathematical part of the post.
Hello again! After a bit of a break, today I thought it’s long time to move on to Warlords of Draenor healing. For anyone who has spent a shred of keeping up, we know that the Beta is now out, and with that many initial theorycrafting formulas are available as well. While I have not had the chance to play the Beta myself, I have taken the time to work with the initial formulas released.
So today, I’d like to go over the topic of Stat Weights. While Theck has been busy working on creating a Holy module for SimulationCraft, today I thought I’d explore some formulation when it comes to figuring out weights. Fortunately, Holy Paladins don’t have terribly complicated spell mechanics, so our formulation-derived results should give us a great ballpark. I’ll also go over some caveats that are not captured in my methods, and let readers mull over them for a bit.
Before I get too deep, please, please remember that Warlords is still in a beta stage, and many of the numbers and specific class mechanics are subject to change. However, my overall methodology for finding stat weights should and will remain applicable to new sets of numbers and even through minor mechanics changes.
1. Your Average Spell: The Methodology
Today, with base healing removed, the average spell HPS (healing per second) looks something like this:
Here, each variable stands for what you might thing, with SPC and CT standing for SP coefficient and cast time. Since we will be looking at percentage gains per stat, in the long run these two values will not actually factor into our weights (a benefit of removing base healing from the equation, by the way!). Note that by considering HPS over simple healing or HPCT (healing per cast time), we are taking into account the benefits we gain from Haste.
To look at which stat, immediately, will give the largest gain we need to first take the derivative (this gives us the average gain for a small amount of each stat added). This is important since if we add too many “units” of stats, we can distort the results, if our gains for each stat happens to be non-linear. Finally, since we are looking at stat weights, we are interested not in the gain per crit chance, but rather per crit rating. So we also need to take that into account for our stat weight formula:
The first term in that formula can be computed from our known formula for crit chance based on crit rating. It follows from taking the amount of rating required for 1% gain in crit chance (110), our attunement, and the 10% base crit we get from innate 5% and the raid buff 5%.
To make the formula derivations simpler, we can actually look only at percentage gains. These will give the same relative weights for comparison which is good enough when it comes to making comparisons. (However, for a program like ReforgeLite, you want the absolute weight, not the percentage gain weights we are deriving now!). Putting our last 3 formulas together, we find the percentage gain weight:
And here we are, a shiny stat weight! We can further modify this formula (for instance, convert C straight to CR), but this should do for most spreadsheet purposes. Finally, we can use this same method to find similar equations for the other stat weights. Note that as you get more of a stat, its percentage gain stat weight will decrease. This is simple to see: Going from 0% critical strike to 25% is an overall percentage gain of 25%. Going from 25% to 50% is a percentage gain of only 20%.
In theory, this means that eventually, if we stack a single stat, the other stat weights will “catch up” – hence we will never want to “over”-stack a specific stat. In practice, often with unequal stat weights the point where our “supreme” stat begins to fall below the other stats will rarely, if ever, happen – especially in only the first tier of raiding.
2. Estimated Stat Weights for Holy Paladins in Warlords of Draenor
With the methodology above, I will assume an average, semi-geared, Level 100 Paladin. Here, I took the Paladin’s stats (from Beta) from gear, added them up, and pretended we were able to “reforge” them all to create equal stats. Finally, I rounded the values to “easier” numbers solely to make our calculations easier – these number changes should have little to no final effect on our relative stat weights that we calculate. From gear, I got the following values:
- Intellect: 3,400
- Spell Power: 1,250 (from Weapon)
- Spirit: 1,150
- Other Stats: 2,800. Divided evenly between Crit, Mastery, Multistrike, Versatility, and Haste to give 560 in each.
- No raid flask and food for now. In Warlords, flasks and food are for secondary stats, and we will eventually flask and food our strongest stat based on these weights (or possibly spirit).
These lend us to values as follows, each of which is the percentage gain for 1 point of each stat, multiplied by 100 for readability:
- Intellect: 2.21 (since we ostensibly cannot gem, flask, or food intellect, this is relatively inconsequential).
- Critical Strike: 0.83
- Haste: 1.17
- Mastery: 0.83
- Multistrike: 0.84
- Versatility: 0.84
3. Unaccounted Factors in Stat Weights
Before we move on, I should restate my first point: The stat weights are for “Your Typical Spell.” However, some spells are indeed atypical, and we should account for that qualitatively, before we take the former stat weights too much for granted:
- Paladins have Infusion of Light, which either grants Haste or effective Mana on Holy Shock criticals. This raises the effective value of crit.
- Holy Shock has a doubled critical strike chance. Since Holy Shock is generally frequently used, this raises the effective value of crit.
- Haste only gives a HPS benefit, while all other stats grant an HPM benefit (healing per mana). Too much Haste stacking may have to be compensated with extra Spirit as well. How much this will factor into our overall stat balance will be 100% dependent on base mana regeneration and our rate of “mana problems” we have.
- Overall “Healing” Paradigms: Generally dependability is preferable to variance, even if output could be sacrificed. In addition, more smaller heals are preferable to few larger heals, due to overhealing. This will generally, all else equal, push stats such as haste, mastery (due to absorption), and versatility ahead of stats such as crit and multistrike.
4. Critical Strike Stacking
Combine our above conclusions with the newly announced “Critical Strike Attunement” and one may figure that we are “supposed” to stack Critical Strike for 6.0. Let’s try this. I have now rearranged our stats to a realistically attainable value:
- Stats are now: 1,100 critical strike, 370 haste, 400 mastery, 710 multistrike, 220 versatility.
- Stat weights are now: 0.79 critical strike, 1.19 haste, 0.84 mastery, 0.83 multistrike, 0.87 versatility.
Critical Strike’s value has already dipped significantly, and this is before even adding a possible crit food and flask.
5. Initial Conclusions
The main conclusion and worry after our foray of math is that critical strike simply isn’t exciting enough, even with the attunement, to justify “stacking.” Since the attunement gives a 5% bonus to stacking single stats, if we are not stacking crit despite the presence of the bonus, then our “bonus” attunement is effectively being wasted. This can push us behind other healers that are able to stack their attunement stat.
Right now, stats such as haste and mastery are looking more attractive despite the critical bonus on Holy Shock – if one recalls back to Mists, even, the effect while desirable is hardly worth seeking by stacking a single, otherwise unattractive stat.
Finally, consider that this is only the first raid tier, and now look to the future, where the return on crit stacking will even further diminish, and one will wonder where we want to put our stats.
6. Suggestions and Reactions
The question then, what can we do about this? What kind of changes should be done?
- Nothing. I disagree with this, because I don’t think that having a state of mediocre results from each stat is very exciting, nor do I think it is the design goal. As I mentioned earlier, having a 5% attunement on crit, and then not wanting to stack crit, is in fact a 5% nerf compared to classes which do stack their attunement stat due to it, on the whole, being desirable.
- Change attunement to a different stat. This could be interesting if we, say, were instead attuned to haste or mastery. I’m not quite sure mastery is the goal, after the absorption debacle of Mists of Pandaria (despite the fact that primarily, discipline priests were the main culprit here). Haste attunement could be interesting, but I worry about us eventually GCD capping, which seems very likely. Given first tier stats, stacking 1,100 haste would yield 31.38% haste. This is enough to reduce the value of Avenging Wrath and Bloodlust. Only a little over double that, 2,390 haste, will full cap us (50%).
- Add more interesting things to critical strike. Perhaps the only thing wrong is that crit doesn’t do enough for us. What can we do to solve this? Some ideas include adding more special effects on criticals. For one example, we could base the “old Daybreak” perk (double Holy Shock after a Flash/Holy Light cast) on critical strike, instead of a base %. Another suggestion would be to add Illumination back, baseline, without the penalty from the glyph. This is something I would heavily support, if only because I don’t find the Illumination glyph very exciting at all. We can also baseline some of the Tier 16 2pc bonus, which gives additional healing to Infusion of Light casts. In fact, thinking some more I believe this would be the best solution, if we could take some of these ideas and bake them in (especially the glyph, I hate seeing useless glyph slots).
I’m interested both in criticism/comments on my methodology used to calculate stat weights (You will see this a lot more), as well as my conclusions regarding our current stat priorities. Any thoughts and suggestions would be more than welcome here!