Destruction Mastery

[Roriwix 4]

Hey I have seen quite a few guides saying mastery is the worst stat for destruction but I can't find any of them showing the math behind the mastery so I can actually make a decision myself - I'm curious to know what the average damage increase on mastery is when it says for example "up to 60% more damage" - does that mean on average 30% increase or on average 10% increase?

[Roriwix 4]

No one knows or?

[Valks 2,373]

On 9/5/2016 at 1:13 AM, Roriwix said:

Hey I have seen quite a few guides saying mastery is the worst stat for destruction but I can't find any of them showing the math behind the mastery so I can actually make a decision myself - I'm curious to know what the average damage increase on mastery is when it says for example "up to 60% more damage" - does that mean on average 30% increase or on average 10% increase?

I personally don't play a warlock, but generally people obtain these conclusions from simulations. You adjust stat weights to produce less/more dps and you can see exactly what is most important and at what breakpoints.

[Smacktard 2]

The way they test stat weights in Simcraft is they set the haste/mastery/crit/vers values all to a control amount (say 4000 each, which is reasonable for mythic 5-man gear levels right now) and run a batch of sims to get a baseline. Then they tell simcraft to run a stat weight comparison, and what it does for that is it takes each stat, runs a few sims, increments a chosen stat by 100, runs a few more, increments it again, runs a few more, etc to determine how much of a DPS gain you get per point of each stat. Then these valuse are all normalized against the DPS increase you get per point of primary stat, meaning that the values you see in guides are relative - primary stats are usually normalized to 1.0 to give a clear relative scale of how strong each secondary stat is. Which is why you'll usually see stuff presented like this:
 

Int: 1.0
Haste: 0.97
Crit: 0.87

etc.

 

The reasoning behind Mastery being considered weak for Destruction is mostly because the bonus damage is random, and calculated on each cast/tick. If it were a mechanic more like the Mage's Incanter's Flow talent that gave clear windows for burst it would be a much stronger stat. From a statistics standpoint, your average bonus should even out over the course of a progression encounter to about half the maximum, but that's not necessarily going to happen because RNG is, well...random. Right now Destruction has the best capacity for burst damage, but also the lowest parse-to-parse consistency from what I understand.

Edited September 6, 2016 by Smacktard

[vaeevictiss 17]

from the class guide 

1. Haste: 1.24
2. Intellect: 1.00
3. Critical Strike: 0.85
4. Versatility: 0.58
5. Mastery: 0.55

so yea, its slightly less than half as strong as haste.

[Roriwix 4]

Right, im not interested in stat weights, im interested in knowing what exactly the math behind the mastery is. How much average damage do I get from 100 rating for example. Ingame it says random amount, but it doesnt state average - obviously simulator knows what the average is, but I'm asking if anyone actually knows "rating/averagedamage%".

Similar to how you need 350 rating to get 1% crit, im asking how much rating do you need to get 1% average damage on mastery. 

The tooltip for mastery requires ~117 rating to increase 1% but that doesn't actually explain the average damage increase.

I've read the class guide, it doesn't explain that part.

[Roriwix 4]

Thank you for your assistance :)

Edited September 8, 2016 by Roriwix
.

[Mikedawg 4]

I think sims are making the assumption that you're using imp+glyph of imp and Mastery doesn't affect your pet damage. If you're using GoSac the stat weights should be different. 

Also in the Destro thread (here or mmo-champ?) there are people saying that there is literally no explanation for the fact that Versatility is above mastery, and that it is mathematically impossible that Versatility is superior. But I didn't read or do any math so I dunno about that.

Edited September 8, 2016 by Mikedawg
grimoire, not glyph

[Roriwix 4]

My sims give me Crit>Vers>Haste>Mastery but thats at 9k haste rating with grimoire. I still aim for crit+haste gear regardless. 

Either way im just looking for factual knowledge about the mastery rather than this second hand statweight stuff.

[Valks 2,373]

So, you essentially want to know how much your DPS will go up every time you get X mastery?

It's not possible to calculate this, at least not for us. You'll need to sit there and do countless sims for average values, as well as with your own gear levels. The value will change every time your mastery goes up or down.

[Roriwix 4]

How hard can it be to understand such a simple thing?

The tooltip for mastery requires ~117 rating to increase 1%

If the average of the randomness on the mastery is half then it would require only 234 rating to get 1% more damage (with sacrifice). That would obviously be OP, so im asking what the average on the mastery is. Is it 1/3, is it 1/4. Simulationcraft obviously knows this so there must be someone who knows, who put it in the program.

[Genu 0]

12 hours ago, Roriwix said:

Simulationcraft obviously knows this so there must be someone who knows, who put it in the program.

Skip to after the next quote to answer the question, this is just a bit of extra information for you to understand SimCraft:

That is actually a misconception. To put it in the programm, you only need to know when and how mastery is applied, not how it does less damage or how it compares to other stats. The SimulationCraft is a Simulation, so all it does is applying various effects in the right order to get a value. When it has all the values (e.g. 1 Fight over 10 Minutes), it creates some fancy output for us to read, like average damage per spell, spell distribution or the damage dealt per effect (like mastery or critical strikes). From the values gathered stat weights will be derived and represented in a readable manner through basic statistics. There is nothing that tells simulationcraft that mastery doesn't average out. It will always average out in infinite iterations. We can't do Infinite iterations, but an arbitrary high amount, which means the sample size is great enough to represent the actual odds with marginal (for people who don't know statistics read "without" to not get confused or misinterpret this) error.

What you see in Stat weights is the normalized average DPS gain from Mastery through hundreds of thousands correctly executed rotations given the specified parameters (Gear, Action Priority List, Talents, Artifact Progress) by Increasing and Decreasing stat values in small steps, then combining these to an average gain and normalizing them for comparison.

12 hours ago, Roriwix said:

How hard can it be to understand such a simple thing?

If the average of the randomness on the mastery is half then it would require only 234 rating to get 1% more damage (with sacrifice). That would obviously be OP, so im asking what the average on the mastery is. Is it 1/3, is it 1/4.

It is not that simple. You don't look deep enough - the rating to % conversion is not the single factor in calculating the damage output. The Average of Mastery gain is half the max value, but it still compares unfavorable to other stats:

Part reason is damage distribution. Not all spells do equal damage and are used equally often - so while in the long term Mastery averages out, you will never achieve this during a 10 minute fight. Spells that are cast more often are more likely to average out, However the spells that would gain the most damage through mastery are cast less often, so they rarely average out and tend to have great swings. SimCraft averages this out through multiple Iterations over multiple Fights. However in a single fight, variance can get you and is more likely to empower the wrong spells (as you are casting the spells that benefit less from mastery more often, see later).

The other part is that you dismiss another thing, while 350 crit are 1% crit and 234 Mastery are 1% average increase, you set both stats as equal damage increase. This is not true - there are talents and abilities that benefit from crit more than flat damage - the second conflagrate from Conflagration of Chaos artifact point, the flat damage increase on Chaos Bolt in combination with the Critical hit damage increase from Chaos instability (which isn't even used in sims yet, that will have great impact) and the generation of Soul Shards - resource gain, even double dipping with Haste through reduction of Conflagrate Cooldown. So there is a reason for crit rating converting lesser to crit%. So in the end even if you gain less % for more stats, going for critical rating simulates more beneficial than going for mastery rating.

If there were no difference in Resource gain and they would average the same, there is one gameplay wise:

With a look at the current Sims, average Chaos bolt damage is ~155k per bolt while Incinerate is ~83k. So Incinerate deals 53% damage of Chaos bolt. A Chaos bolt with the full 60% damage boost means you get more than a free incinerate! However, you cast Incinerate twice as often as Chaos bolt so you are more likely to average out on +30% on Incinerate, while Chaos bolt will swing more wildly from 20%-40%.

Mastery also stacks with crits meaning empowering a non-Crit Incinerate or a critical Incinerate with 60% is a huge difference. That obviously doesn't happen with Chaos Bolt, but it does with all the other Spells, increasing randomness in a single fight through decreasing your chances of empowering the right hit with decreasing critical hit chance.

This implies that Mastery is stronger on Chaos bolt than the other spells, as it always critical strikes and has the most constant damage. However, Chaos Bolts are about 15% of the spells you cast during a fight, meaning 85% of the time your mastery has a lesser benefit (read: more randomness, not less average damage).

On average it is still the same damage gain, but you want to have constant dps to plan for bursting down targets and be able to plan your damage every fight instead of having huge dps swings. You simply don't want to wipe the Raid because you had RNG and couldn't kill your add.

So the TLDR:

- The Average of Mastery gain is half the max value

But there are Problems with Mastery:

- Mastery doesn't affect Resource gain

- Mastery does not empower Talents / Artifact Traits

These upper two are the reasons why mastery compares bad to other stats. Then there are some gameplay problems:

- Mastery scales with Crit (empowering critical strikes nets more damage), increasing randomness even further - see next point)

- Mastery damage  cannot be planned is less reliable than Crit (as Mastery dips from crits but not vice versa) or Haste (constant faster casting and resource gain), there is continuity in performance needed in raiding

For Mikedawg:

To the Versatility Problem: I cannot deliver more insight than Simulationcraft does. If there are some people saying it shouldn't be better, there are 2 possibilies: They either calculated wrong (skewed view, sample size not high enough to represent actual odds, messed up the orders in where effects are applied) or they discovered a flaw in SimC (most likely effects applied in the wrong order).

As with any program / tool i only can trust the results if i trust that it works correctly. Maybe Mastery is better than Versatility or maybe "people" without known background on the forums don't know what they are talking about. The people distributing to SimulationCraft have proven to be a reliable Source and many where hired by Blizzard to do that exact same thing there.

Of course there can and will be Errors and it is good to discuss these things and have people look at it, but when i use a tool, i don't question the results if i know that i used it properly. (And didn't use it to verify that it works properly)

 

 

Edited September 12, 2016 by Genu
Tried to clarify some sentences.

[Roriwix 4]

Quote

That is actually a misconception. To put it in the programm, you only need to know when and how mastery is applied, 

And that is literally what I'm asking for - how is mastery is applied xd

Quote

The Average of Mastery gain is half the max value, but it still compares unfavorable to other stats:

Well today I found it that it is 1/3 of the value in Blizzards metric - I got a mastery buff that gives 25% mastery and it increased the tooltip by 75%

Quote

 the rating to % conversion is not the single factor in calculating the damage output. 

Ofc not - but when my question is to found out what the tooltip actually means then I'm not really interested in the overall output.

Quote

Maybe Mastery is better than Versatility or maybe "people" without known background on the forums don't know what they are talking about. 

Again - I'm not debating whether mastery is better stat - just trying to find an answer to a simple question - one that noone could answer because they only regurgitate end results. I found my answer myself tho so its all good.

Edited September 13, 2016 by Roriwix

[Genu 0]

6 hours ago, Roriwix said:

Well today I found it that it is 1/3 of the value in Blizzards metric - I got a mastery buff that gives 25% mastery and it increased the tooltip by 75%

Do you care to cite your source to answer this question for everyone and explain how theses two things are related?

Or was your first question was actually about how mastery damage bonus scales with mastery rating increase and not about the average damage that mastery provides?

Well regarding to the formula, later on i will give you the summary of how mastery is applied in the current SimCraft as soon as i am through the code.

 

Edited September 13, 2016 by Genu

[Valks 2,373]

3 hours ago, Genu said:

Or was your first question was actually about how mastery damage bonus scales with mastery rating increase and not about the average damage that mastery provides?

Judging by how badly this thread has spiralled and how many different answers have been given, I don't think anyone knows what the question is except the OP. It must be too simple for us to understand.

I'll keep this thread open rather than locking it, but in all honesty, I don't actually know what the whole point of it was, nor what the conclusion the OP came to was. Did we actually find the answer? Did anyone even find the question?

[Genu 0]

Thank you for keeping it open, this way i can at least answer the part how mastery is applied and the damage calculated.

Mastery is a spell multiplier that is applied last after the damage of the spell/dot tick has been calculated (except for mastery at this point, obviously). The bonus value is chosen through a random number generator by uniform distribution in range. The mean of uniform distribution is 1/2 of the range. The range of the mastery is determined by 0 to 1, which translates to your maximum mastery bonus value. (e.g. The rng gives you a 0.5 you get half of your maximum mastery bonus value as damage bonus). For more information you should look into how mastery rating converts to your maximum mastery bonus value.

As result of uniform distribution, the average damage increase through mastery that is to be expected is half of the maximum Mastery bonus value (because uniform distribution is not skewed). E.g. If your tooltip states that your mastery increases your damage "up to 50%", you can estimate that mastery will account for 25% of your overall damage on average.

Since mastery is a spell multiplier, it is not listed as a seperate damage source on the SimulationCraft results page but included in the spells damage. Including it in the damage of spells is a design choice to make damage per execution time comparable, which is a major indicator for spell priority.

I hope that clarifies how mastery works.

While mastery is a great damage increase, see the issues stated the large post above.

//Complement:

There is another issue with mastery: As it is applied when damage is dealt, it does not apply to the Roaring Blaze talent. That means with up to 50% mastery bonus, if you have 5 Immolate Ticks for 10K each remaining, they would combine to 50K for internal tracking since mastery is not applied yet, but deal 60K (12.5K per Tick on average with mastery applied) when they actually deal damage . Roaring Blaze only sees 50K damage to apply to. (The empowered Ticks after roaring Blaze was applied still trigger the mastery).

That is the reason why Versatility currently is better than mastery, since it does apply to the internal damage tracked and scales better with our most significant source of damage.

Edited September 13, 2016 by Genu
corrected some spelling errors. Misinterpretable sidenote removed.

[Roriwix 4]

3 hours ago, Blainie said:

Judging by how badly this thread has spiralled and how many different answers have been given, I don't think anyone knows what the question is except the OP. It must be too simple for us to understand.

I'll keep this thread open rather than locking it, but in all honesty, I don't actually know what the whole point of it was, nor what the conclusion the OP came to was. Did we actually find the answer? Did anyone even find the question?

I'm sorry if you feel the thread has spiralled out of control. I asked a question and people answered a ton of different questions - there is a lot of useful information in this thread and if I seem hostile to you then thats just my writing style, I can't be bothered to sugarcoat things, I'm not actually angry at anyone. I got the information I wanted myself so and I'm sure others can use all the other information that is in this thread just fine so I think you should leave it open so others can ask questions about the information that was given if they want.

I got a buff in-game that gave me 25% mastery - that increased the tooltip by 75% thus I assume that in-game the mastery averages out 1/3. But its good to know that simulationcraft runs with 1/2 and accounts for all the losses that occur by not being applied to certain spells.

TL;DR I knew that mastery was shit, was just looking to know how shit it was and I found that answer.

[Genu 0]

The increase of the mastery damage bonus in the tool tip through mastery rating has nothing to do with the actual values rolled. your assumption that it averages out at 1/3 is simply wrong, since the value at which mastery bonus caps out does not have an effect on the average expected value of the function that determines how much of your maximum mastery bonus is applied. You mixed things up.