In almost all cases, our mathematical models produce similar results to advice you will find elsewhere on the internet – as you would expect since we are all playing the same game!
Protection warrior mastery is an exception though… no matter how we run the numbers, it always comes out as a significantly better stat than other sources say.
Is there an error in our calculations? Are there variables being taken into account by others that a calculation can’t capture?
Usually if we go over our code and calculations enough times, compare it to logs, and so on… we can converge on a similar result to everyone else. It’s a bit frustrating that we can’t in this case, so we’re looking for some feedback to try and figure out what is causing the difference.
General Stat Priority
To start, let’s summarize the general position on mastery (and other stats) that you’ll find out there:
Wowhead
To paraphrase: versatility and critical strike are of similar value, haste is slightly better, mastery is slightly worse, but all secondary stats are pretty good. The reason that mastery comes out below the others is because it does not improve your damage reduction against unblockable sources, and it grants the least damage output.
Icy Veins
It is essentially identical to the wowhead guide in both conclusions and phrasing.
Method
Again, the same stat priority. It is a little more negative about mastery, saying that because of stats like Shield Specialization the defensive part of mastery is a bit redundant.
Analysis
So let’s dig into this, focusing on comparing mastery to versatility. We’ll look at defense, then offense.
As an aside: why are we ignoring critical strike and haste? Because they are generally weaker than mastery and versatility for defense, and generally better for offense. The preference for these stats comes from an overall preference in the community for offensive-leaning builds for tanks. Keep this in mind – you must move the slider on our site closer to the DPS side to get builds that are similar to the generic recommendations elsewhere. We are one of the few resources that gives you an option for a purely defensive build, and it will look very different.
Haste is also heavily preferred for “feel” more than any numerical benefit. Haste will always be worse on paper than people feel it is in-game, because people generally like the shorter cooldowns and faster gameplay. Just note that this faster gameplay will NOT necessarily make you tougher or do more damage than balancing haste with other stats. We adjust for this preference in our optimizer to an extent. We think that some people go overboard on haste to the deteriment of raw performance, so we make this adjustment subtle enough to get an amount of haste that still feels pretty good but is in balance with other stats.
Also note that haste’s value can depend heavily on your chosen talents. For example, some combinations require a lot more haste to keep high shield block uptime, and thus haste will be preferred more. Other combinations require very little haste to keep shield block up, and thus lower values will be recommended. And perhaps counterintuitively, a talent like Into the Fray generally makes haste better as it multiplies with your haste from gear.
Defense
Versatility
780 versatility rating will reduce your damage taken by 0.5%, and this applies to every type of damage that you take. It multiplies with every other form of damage reduction so there isn’t any weirdness complicating things.
If we convert to an equivalent required to get 1% damage reduction (DR):
1560 versatility rating = 1% DR
780 versatility also increases all healing and shielding (most notably ignore pain) by 1%.
Mastery
Mastery has three effects that can impact damage reduction:
700 mastery rating increase block chance by 0.5% (before DR)
700 mastery rating increases critical block chance by 1.5%
700 mastery rating increases attack power by 1%
The block chance part isn’t very important – we would agree that it is usually redundant due to high uptime on shield block, so we’ll just ignore it.
The attack power part can more or less be considered a 1% increase to all healing and shielding. Ignore pain scales with attack power, and the other largest source of healing for a warrior is usually leech which scales off your damage, almost all of which scales with attack power. You might have some trinket effects that don’t scale with attack power, but the healing part is usually somewhat small for protection warriors so this is an OK approximation for comparison purposes.
The critical block chance part is very valuable. It is worth looking briefly at how blocks work to understand its value.
Blocks are functionally similar to armor. Your shield has some large-ish number on it, like 92757 for an ilvl678 shield (max level season 2 gear). This can be multiplied by a few talents, like Shield Specialization and Brace for Impact, so your final block value would be more like 132271 most of the time. This will decrease the damage of a blockable attack by ~46% against a raid boss in a mythic raid (there is a formula for this and some constants that can be extracted from the game, we can post the full details if anyone is interested).
NOTE: The effectiveness of block (and armor) goes down at higher difficulties. There is a constant associated to each difficulty that increases the armor/block constant (and thus decreases the effectiveness of those stats). For example, in a heroic raid this example would decrease damage by ~48% instead. We will be talking about mythic raid difficulty, which is the worst case value of mastery for a protection warrior.
A critical block simply doubles this amount: so 46% becomes 92%… but there is another consideration: no single source of DR can exceed 85%. So a critical block is capped out at an 85% damage reduction in this example. It is not difficult to cap out your critical blocks at 85% reduction (and it’s not hard to go to a tanking dummy and see this in action by throwing up shield block and looking at your combat log for a few hits). (Side note… for this reason we find the Martial Expert talent in the Colossus tree a bit weird… it’s only useful at low gear levels where you aren’t already capping out the size of your critical blocks).
So in this example, we can use a little algebra to convert our 700 mastery into an equivalent damage reduction percent to compare it to versatility:
regular block = 46% DR = 1 - 0.46 = 0.54 multiplier on damage taken
crit block = 85% DR = 1 - 0.85 = 0.15 multiplier on damage taken
If we are assuming a near-100% block chance due to shield block, we are assuming that we would have at least done a regular block on any blockable attack. So the value of increasing our critical block chance is only the gain over a regular block, not the full value of a critical block. We can determine this by thinking of a critical block as an extra DR multiplier applied on top of a regular block:
0.15 = 0.54 * (1 - x)
x = 1 - 0.15 / 0.54 = 0.72
Thus a crit block is equivalent to applying another 0.72 or 72% DR to that attack.
So now let’s get the amount of mastery rating required to get 1% DR on average over all attacks via critical blocks:
700 rating = 1.5% crit block
1.5% crit block = 0.72 * 1.5 = 1.08% DR
648 mastery rating = 1% DR
But… not all damage can be blocked, and the amount can be highly variable based on content. Mythic+ is very heavy on blockable damage. Raids are lower. But if you look at any random log… it is rare that a protection warrior’s damage taken is less than ~40% from melee attacks. That’s damage TAKEN… and a warrior’s DR on blockable attacks is very high compared to other kinds of attacks, so even in a raid it is very rare for less than ~60% of raw damage to be blockable. We’ll use 60% as a decent worst case estimate:
648 mastery rating = 0.6% DR (only affects 60% of incoming damage)
1080 mastery rating = 1% DR on average over a worst-case fight for blockable damage
Offense
The comparison of mastery to versatility is interesting. Almost all damage scales with attack power. Mastery is a straight multiplier on attack power, so generally speaking 700 mastery rating = 1% damage gain, and 780 versatility = 1% damage gain. Where things can shift in favor of versatility is direct damage effects from gear like trinkets – those do not scale based on attack power and thus mastery has no impact. If these are enough of your total damage, then versatility overtakes mastery by a small amount.
In practice, mastery and versatility are extremely close in value for offense, with versatility being a little “safer” as a damage stat because you don’t need to worry about it suddenly losing some value if you change your trinkets.
Questions
OK that was a lot more detail than anticipated, but it raises one big question: why does almost everyone rank mastery as the weakest stat? By our estimations, it is significantly better than versatility for defense, and approximately equal or just slightly worse for offense.
If we just go ahead and say mastery is slightly worse in all real world scenarios for offense, even if you are going for a higher offense build… mastery’s superior efficiency as a defensive stat would probably make it a better use of budget than versatility in all but the most skewed builds (i.e. you could get just a bit of mastery to cover your defensive needs, and gear more heavily for crit and haste to have better overall damage without being too squishy).
What could be causing the difference in results here?
-
Is there a flaw in our calculations? Let us know if you see one!
-
Is the fact that part of mastery only applies to blockable attacks perceived as a big downside, to the point that it massively devalues it as a stat?
-
Similar to 2: is the fact that crit blocks are a chance and thus feel a bit like dodges and parries making it unpopular because it is unreliable?
My suspicion is that it is some combination of 2 and 3. People prefer the guarantee of versatility, even if over the course of a fight it is significantly less reduction than an equivalent amount of mastery.
Following up on the calculations above, in a worst-case scenario for mastery it would bring ~40% more total damage reduction than an equivalent amount of versatility. In Mythic+ it would be significantly better than that.
The more random nature of critical blocks and the restriction of part of mastery’s impact to only physical damage should certainly penalize its value a bit. But 40% is kind of a lot of “free” damage reduction to leave on the table… are we perhaps being excessively punitive with our evaluation of mastery? Especially in a Mythic+ setting we’re talking about mastery being 60% or more effective for overall damage reduction than versatility, point-for-point. Should we be favoring mastery more heavily at least for Mythic+?
Please let us know your thoughts!