Remove Advertisements

Avoidance diminishing return

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Avoidance diminishing return

Postby Paxen » Mon Sep 17, 2012 1:37 am

Maybe it's a silly question, and the point isn't that important since avoidance is inherently the worst form of damage reduction (compared to armour or other "always on" reductions).

Diminishing return on avoidance is generally regarded as placing a "soft cap" on avoidance, ie as you get more avoidance it becomes less useful.

However, avoidance has inherent increasing returns, in the same way all the systems where you add percentage point have - 1% more avoidance is twice as valuable when you have 50% compared to if you have 0%. Blizzard added diminishing returns in WotLK to counteract this.

What I've never seen is a calculation that determines that the diminishing return on avoidance rating is stronger than the inherent increasing return. For other mechanics, that isn't so - armor is regarded as linear, and according to an article on ArenaJunkies Resilience has increasing returns even with a diminishing returns mechanic on Resilience rating. Maybe Theck ran those numbers a long time ago, and I missed it?
Paxen
 
Posts: 609
Joined: Mon Aug 17, 2009 3:38 am

Re: Avoidance diminishing return

Postby theckhd » Mon Sep 17, 2012 9:51 am

I think what you're looking for is mostly answered in these two blog posts:
http://sacredduty.net/2012/08/14/mop-be ... ts-b15961/
http://sacredduty.net/2012/08/17/mop-be ... ts-b15983/

But to address your question specifically: You're asking whether the damage reduction of X avoidance rating goes up or down as you stack more and more of it. Implicit in that question is that you're interested in relative damage reduction, not absolute damage reduction. Absolute damage reduction obviously goes down due to diminishing returns (1% dodge is always going to reduce more absolute damage than 0.9% dodge).

The quantity in particular that you're interested in is dAd/(1-A). dAd is the amount of post-DR dodge you get from 1% pre-DR dodge, while A is your overall avoidance. dAd is obviously dependent on the amount of dodge you have, as is A. If we fix parry at a certain value (say, 10%), and plot dAd/(1-A) as a function of post-DR dodge Ad (with dAd given in the blog post, and A=A0+Ad), we get this plot:
Image

A similar calculation can be done for parry:
Image

The heavily relaxed parry cap causes it to have a slight positive scaling of the sort you're describing, bumping its value by around 3% for every 5% additional parry gained.

Before you decide to interpret that as "parry is a king stat" though, remember that all of the other stats also gain in relative damage prevention by the same factor of 1/(1-A). This is why I analyzed absolute damage reduction in the second blog post - it's simpler, and if there's a crossover point between, say, parry and mastery, that same crossover point exists at the same place on the associated relative damage reduction plot.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7753
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Avoidance diminishing return

Postby Paxen » Mon Sep 17, 2012 2:15 pm

theckhd wrote:But to address your question specifically: You're asking whether the damage reduction of X avoidance rating goes up or down as you stack more and more of it. Implicit in that question is that you're interested in relative damage reduction, not absolute damage reduction. Absolute damage reduction obviously goes down due to diminishing returns (1% dodge is always going to reduce more absolute damage than 0.9% dodge).


Yeah, I thought the Time to Live-paradigm pretty much states that as tanks we're mostly interested in relative damage reduction?

The quantity in particular that you're interested in is dAd/(1-A).


The heavily relaxed parry cap causes it to have a slight positive scaling of the sort you're describing, bumping its value by around 3% for every 5% additional parry gained.


So Parry is in the same boat as Resilience, there is diminishing returns, but not enough to break the inherent increase if you value relative damage reduction. Is it wrong to say that there is no more DR on Parry than there is on Armor?

Is this flipped for Druids and Monks? That is, since they get Dodge from their primary stat instead of Parry, is the Dodge cap for these classes inflated in the same way the Parry cap is for plate tanks?

And does anybody happen to know of any solid theorycrafting on Monk Brewmaster stats. I've only seen one guy on the official forums, and I'm not sure if his work is solid or not. I take it Parry rating is out, even if Blizzard did at one point say they wanted Brewmasters to focus on Parry.

Before you decide to interpret that as "parry is a king stat" though, remember that all of the other stats also gain in relative damage prevention by the same factor of 1/(1-A). This is why I analyzed absolute damage reduction in the second blog post - it's simpler, and if there's a crossover point between, say, parry and mastery, that same crossover point exists at the same place on the associated relative damage reduction plot.


Am I understanding the second post right when I think that Parry has synergy only with itself (and dodge), while Mastery has synergy with haste/hit/exp as it is linked to active mitigation? And that this is caused, at least partially, by block being moved to a second roll, thereby breaking the "add percentage" behavior that causes avoidance to increase the gains in relative damage reduction?

Finally, something of a thought experiment: Say I have a paladin that's locked into at least 60% avoidance, but only 8% Mastery (base). Then Parry would be the thing to stack, disregarding RNG reliance. But as we can dump a lot of that avoidance through reforging or switching to Mastery gear, Mastery (and the dps stats) becomes the better option.
Paxen
 
Posts: 609
Joined: Mon Aug 17, 2009 3:38 am

Re: Avoidance diminishing return

Postby theckhd » Mon Sep 17, 2012 2:42 pm

Paxen wrote:Yeah, I thought the Time to Live-paradigm pretty much states that as tanks we're mostly interested in relative damage reduction?

That's one metric to consider, but it's rare that a tank dies from long-term sustained damage. If you're going to die in current content, it's going to be from a 5- or 6-second burst window.

Paxen wrote:So Parry is in the same boat as Resilience, there is diminishing returns, but not enough to break the inherent increase if you value relative damage reduction. Is it wrong to say that there is no more DR on Parry than there is on Armor?

Yes, I'd say that statement is incorrect. There's obviously diminishing returns on parry, because you just noted there is. The proper way to state what you're getting at is probably that despite diminishing returns, parry's incremental value increases as you stack more of it.

Though, again, I think even that statement is fairly misleading. As you stack more parry, all of the other stats get better too. This effect isn't happening in a vacuum.

Paxen wrote:Is this flipped for Druids and Monks? That is, since they get Dodge from their primary stat instead of Parry, is the Dodge cap for these classes inflated in the same way the Parry cap is for plate tanks?

Possibly. I only have early estimates from beta, and I haven't seen any follow-up work, so I don't have accurate or current estimates of the dodge caps. Last I checked it was ~150 for druids and >460 for monks. If those values are still in the right ballpark, then I'd expect that it's not true for druids, but is true for monks.


Paxen wrote:Am I understanding the second post right when I think that Parry has synergy only with itself (and dodge), while Mastery has synergy with haste/hit/exp as it is linked to active mitigation? And that this is caused, at least partially, by block being moved to a second roll, thereby breaking the "add percentage" behavior that causes avoidance to increase the gains in relative damage reduction?

More or less, yes. Parry and Dodge have synergy with each other, and Hit/Exp/Mastery/Haste have synergy with each other.

Paxen wrote:Finally, something of a thought experiment: Say I have a paladin that's locked into at least 60% avoidance, but only 8% Mastery (base). Then Parry would be the thing to stack, disregarding RNG reliance. But as we can dump a lot of that avoidance through reforging or switching to Mastery gear, Mastery (and the dps stats) becomes the better option.

I think 60% is a bit of a high estimate, given that entry-level gear sets are only capable of managing about 20% total avoidance. But yes, parry is the one stat which, if stacked, doesn't seem to diminish below the others.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7753
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA

Re: Avoidance diminishing return

Postby Paxen » Mon Sep 17, 2012 3:42 pm

theckhd wrote:
Paxen wrote:Is it wrong to say that there is no more DR on Parry than there is on Armor?

Yes, I'd say that statement is incorrect. There's obviously diminishing returns on parry, because you just noted there is. The proper way to state what you're getting at is probably that despite diminishing returns, parry's incremental value increases as you stack more of it.


Yeah, you're right. There is diminishing returns. But doesn't that apply to armor as well? You get less and less damage reduction per point of armor the more you have?

theckhd wrote:If those values are still in the right ballpark, then I'd expect that it's not true for druids, but is true for monks.


Thanks. I guess Monks really aren't Parry tanks after all.


theckhd wrote:More or less, yes. Parry and Dodge have synergy with each other, and Hit/Exp/Mastery/Haste have synergy with each other.

theckhd wrote:But yes, parry is the one stat which, if stacked, doesn't seem to diminish below the others.


To me it seems that you should either gem/reforge for a mix of Hit/Exp/Haste/Mastery, or go all out for Parry. Is that too simplistic? (My gut instinct would be to dump Parry.) Or is the stat weights high enough for Parry that I should still reforge away Hit/Exp from an item with Parry/Hit(Exp)?
Paxen
 
Posts: 609
Joined: Mon Aug 17, 2009 3:38 am

Re: Avoidance diminishing return

Postby theckhd » Mon Sep 17, 2012 4:25 pm

Paxen wrote:Yeah, you're right. There is diminishing returns. But doesn't that apply to armor as well? You get less and less damage reduction per point of armor the more you have?

Yes. But the DR function used for armor is different, and it's balanced such that it gives a flatter relative DR benefit.

Paxen wrote:To me it seems that you should either gem/reforge for a mix of Hit/Exp/Haste/Mastery, or go all out for Parry. Is that too simplistic? (My gut instinct would be to dump Parry.) Or is the stat weights high enough for Parry that I should still reforge away Hit/Exp from an item with Parry/Hit(Exp)?

It's a bit more complicated than that. If you go the dodge/parry route, you'll still want to keep dodge and parry balanced to minimize DR losses. If you go the hit/exp/haste/mastery route, then you won't really care what the stat weights are, because you're already choosing to take smoothness over raw damage reduction.
"Theck, Bringer of Numbers and Pounding Headaches," courtesy of Grehn|Skipjack.
MATLAB 5.x, Simcraft 6.x, Call to Arms 6.0, Talent Spec & Glyph Guide 5.x, Blog: Sacred Duty
User avatar
theckhd
Moderator
 
Posts: 7753
Joined: Thu Jul 31, 2008 3:06 pm
Location: Harrisburg, PA


Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: No registered users and 1 guest


Remove Advertisements

Who is online

In total there is 1 user online :: 0 registered, 0 hidden and 1 guest (based on users active over the past 5 minutes)
Most users ever online was 380 on Tue Oct 14, 2008 6:28 pm

Users browsing this forum: No registered users and 1 guest
?php } else { ?