Dodge/Parry cap before DR

jere wrote:If you really wanted to get technical, dodge rating isn't affected by DR. The conversion from dodge rating to dodge percent is unaffected by DR. It's the dodge % after adding dodge from rating and agility and defense that is then affected by DR. But we honestly shouldn't be making those distinctions. We should be saying things like "the avoidance from defense is subject to DR" and "all avoidance minus those from % talents and naked stats is subject to DR, whether it come from dodge rating, agility, or defense".

As this is an informational site, we try to keep the quality of the information as high as possible. Hence we 'fix' tiny mistakes like these as to make sure there is no way to interpret them wrong. If someone claims that defense suffers from diminishing returns, that needs to be corrected. Wording is key, and I agree with your way of saying it.

Also, the conversion from dodge rating to dodge percent is exactly where the DR takes place. Not before, and not after. Simply because some sources for dodge percent are not subject to DR while others are.

trellian wrote:As this is an informational site, we try to keep the quality of the information as high as possible. Hence we 'fix' tiny mistakes like these as to make sure there is no way to interpret them wrong. If someone claims that defense suffers from diminishing returns, that needs to be corrected. Wording is key, and I agree with your way of saying it.

I am aware of this site's mission. I have been here for quite a long time, possibly longer than you think. However, his response wasn't a fix, it was a nitpick. In the part of my post he responded to, I specifically mentioned that the avoidance from defense was what was subject to DR. What I said was correct and complete, especially considering the context of the question that Aragao was asking. Badgermonkey had him thinking that stuff from defense wasn't affected by DR. I said the avoidance from defense is affected by DR (unless you are saying that that phrase is incorrect).

trellian wrote:Also, the conversion from dodge rating to dodge percent is exactly where the DR takes place. Not before, and not after. Simply because some sources for dodge percent are not subject to DR while others are.

No, it happens after. The DR equation is not linear and does not follow the superposition principle. If you did the DR equation at conversion, you would get the wrong result as y(x1+x2+x3) != y(x1) + y(x2) + y(x3). DR happens after conversion. If you could add the results from various DR equations, then it could happen at conversion, but this is not the case as the results added together do not provide the same result as adding avoidance together and then running the DR equation.

The correct process for processing DR is:

Convert Dodge Rating to Dodge Percent
Convert Agility to Dodge Percent
Convert Defense Rating/Skill to Dodge Percent (doesn't make a difference which since this conversion is linear)
Sum the results of the last 3 operations
Subtract out dodge % that is not on DR
Run DR formula.
Add in the dodge % you subtracted out

By the time the DR equation is run, conversion has already happened.

I think the original poster is looking for a number that maximizes avoidance and minimizes diminished returns. A sweet spot, so to speak. Obviously there is no magic number for that.

To minimize the diminished return against percentage dodge and percentage parry, you would want them both to equal the same percentage (i.e. 25% and 25%)

In my experience, it's a bit tough to get that much parry, but ideally that would minimize the total amount of return diminished against the two stats.

To be honest though, you shouldn't worry too much about diminished returns, they're a necessary evil and calculate them into all your gear decisions. (You have tankadin2 I hope. If not, get it.)

If you want to know if you're ready to maintank high-end content, make a post in the Gear Discussions and Advice section and link your armory. The tankadins here are knowledgable and helpful, you'll have answers and advice within minutes.

Good luck!

jere wrote:The correct process for processing DR is:

Convert Dodge Rating to Dodge Percent
Convert Agility to Dodge Percent
Convert Defense Rating/Skill to Dodge Percent (doesn't make a difference which since this conversion is linear)
Sum the results of the last 3 operations
Subtract out dodge % that is not on DR (talents, naked stats, etc.)
Run DR formula.
Add in the dodge % you subtracted out

By the time the DR equation is run, conversion has already happened.

I hadn't thought of it like that. I would've assumed that they internally would convert all sources of dodge into dodge rating, add those all up, run dr, then convert to dodge percent and add extras... this purely as you would get a more acurate end result. I suppose your way is just as feasable.

How would that be more accurate? On the surface that method is no different in accuracy at all. Also, why add the extra step to covert to dodge rating? The DR equation takes percentages as inputs already.

I am not asking those questions to be mean by the way. They just don't make sense to me, so I think I am missing something in your logic is all.

For those replying to my post, I was tired, hence why it's a half thought out post.

I did of course mean that Defense Rating doesn't suffer diminishing returns, by virtue of it providing a flat number of defense. The avoidances gained from this are subject to it (how OP would it be otherwise ). Block rating isn't, and I don't know why I was talking about armor when I don't have a clue about the damage reduction thing. I probably just meant 10 armor is always 10 armor.

/steps away from the thread.

Armor is subject to diminishing returns in that 100 more armor does not always give the same % reduction in damage taken depending on your total, but, 100 armor will always increase your TTL by the same amount.

DR% = Armor / (Armor + (467.5 * AttackerLevel - 22167.5))

Let's assume AttackerLevel = 83:
DR% = Armor / (Armor + 16635)

Also, let's assume you have exactly 10000 hit points.



Now let's assume you're naked:
DR% = 0 / 16635 = 0%

Since your armor doesn't reduce any damage, a boss would need to hit you for 10000 damage to kill you.

For the sake of simplicity, you found a piece of chest armor with 1848,33 Armor on it.
DR% = 1848,33 / (1848,33 + 16635) = 10%

Suddenly, the boss would need to hit you for 11111,11 damage to kill you.



Now another time. Let's assume you have a couple of items that total up to 16635 armor.
DR% = 16635 / (16635 + 16635) = 50%

Since you reduce the damage in half, the boss needs to hit you for 20000 damage to kill you.

For the sake of simplicity, you found a piece of chest armor that increases your total armor value by 1848,33 Armor.
DR% = 18483,33 / (18483,33 + 16635) = 52.63%

Now the boss needs to hit you for 21111,11 damage.



So no matter how many armor you're already wearing, equipping X amount of armor raises your effective health by Y.

It's also interesting to note that the more stamina you have, the more effective armor is and vice versa, which is why you want to raise both.

JRichland wrote:To minimize the diminished return against percentage dodge and percentage parry, you would want them both to equal the same percentage (i.e. 25% and 25%)

This isn't quite right either, if you meant it the same way I'm reading it. Dodge and Parry use different diminishing returns formulas, so keeping them equal isn't actually ideal.

There's several old, old threads on this from beta and early Wrath, but this one is probably the clearest.

harlekin wrote:1) Stack ~700 defense rating (uncrittable)
2) Stack defense rating and block rating until 102.4% (Paladins only, maybe warriors)
3) Stack dodge rating until dodge rating = defense rating
4) Stack parry rating until 200 rating (assuming dodge = defense = ~700)
5) Add dodge, defense, and parry ratings at about a 10:10:2 ratio (at very high values the parry contribution will grow somewhat)

harlekin wrote:Parry rating is still a bad choice, as the rating conversion and the diminishing returns are both much worse than dodge and defense. The reason that defense rating seems to suffer so much is that the miss DR and the poor parry scaling both diminish the balanced rise. It takes extremely large amounts of ratings to make them better than dodge.

jere wrote:I am aware of this site's mission. I have been here for quite a long time, possibly longer than you think. However, his response wasn't a fix, it was a nitpick.

It was a fix because some people were already slapping their chests in confusion. He was merely illuminating for those confused by previous statements why people like you were starting stupid arguments about semantics. The net result is that the people who want to bicker about semantics get to, and hopefully the people who are here to learn are less confused by these stupid-ass arguments.

The post above mine contains all the information the OP needs to answer his question.

/thread

theckhd wrote:

JRichland wrote:To minimize the diminished return against percentage dodge and percentage parry, you would want them both to equal the same percentage (i.e. 25% and 25%)

This isn't quite right either, if you meant it the same way I'm reading it. Dodge and Parry use different diminishing returns formulas, so keeping them equal isn't actually ideal.

One other thing you can look at that is interesting from a math perspective is you can determine the rates at which each stat diminishes and find the point where those values are equal. That would be a point where stacking one is better than another. I actually worked the math out on that back a bit ago when Invis had his DR question thread up, though I don't think I posted the math there. I could probably still do that if you are interested at all Theck.

It's a slippery slope, though, since parry diminishes so quickly. The moment you find a crossover point, you can drop below it again simply by adding parry. You almost have to bake in some sort of buffer space to avoid that.

jere wrote:One other thing you can look at that is interesting from a math perspective is you can determine the rates at which each stat diminishes and find the point where those values are equal. That would be a point where stacking one is better than another. I actually worked the math out on that back a bit ago when Invis had his DR question thread up, though I don't think I posted the math there. I could probably still do that if you are interested at all Theck.

It's a slippery slope, though, since parry diminishes so quickly. The moment you find a crossover point, you can drop below it again simply by adding parry. You almost have to bake in some sort of buffer space to avoid that.

The formulas are simple enough that it might be worth showing the full derivation just as a reference for others. If you have the time to repost your old calculations, I think it'd be a convenient resource. I vaguely remember a tankspot post that does this already, but my memory might be off.

I like harlekin's approach personally. As you said, the balance changes as soon as you add points in any one of the three stats. Thus, the most straightforward way is to approach it by asking:

Given that I have X dodge, Y parry, and Z defense, in what ratio do I add points to maximize avoidance gains?

If you're close to balanced, then it should give you the appropriate ratio (10:10:2, or whatever it ends up being for current gear levels). If there's a large imbalance, the answer to that question should come out solidly in favor of one stat (i.e. 10:0:0 if dodge is better by some minimum margin).

Again, the math is easy enough that a Java applet could be written to generate the results. I'd have to dig up notes from 7 or 8 years ago to remember how to code a working Java applet though.

This was a first pass at a document I uploaded a bit ago:
http://sites.google.com/site/jerebearst ... stants.pdf

The basic gist is that you can take the derivative of the DR equations to find "how quickly" the output of the DR equations such that you basically have dy/dx, where y is the avoidance value after DR and x is the avoidance value before DR.

I expressed my results in that document as functions of your character sheet avoidance, with one caviat: you need to subtract out talents and naked stats (they aren't affected by DR obviously).

Basically all you would need to do is take your character sheet %'s (the after DR ones) and subtract out 10% parry (talents/base) and around 10% dodge (depends on race...humans are 10% -- talents/naked stats/base). Once that is done, simply plug in either your parry or your dodge into the equation and see what the other one comes out too (I typically plug in parry to see what my dodge needs to be at for crossover). Mind you, the resultant is also minus talents/naked stats/base, so adjust accordingly.

EDIT: It's a PDF format since I can never get equation editor to show up in google sites word documents.

I'm wondering whether those graphs linked take into account that once you get enough rating for 102.4% avoidance, Def loses some usefulness since the block portion of it is wasted.

I'm currently sat at:
743 defense rating
583 dodge rating
267 parry rating
43 block rating

So with buffs I hit the golden 102.4%, but I'm trying to work out at what point, if any, defense or parry is going to equal or overtake dodge since I'm starting to suffer heavily from the DR on dodge.

Well, the effects of block really have no impact on the topic though. Block isn't subject to DR and isn't really avoidance so it doesn't fall into this kind of comparison.

That isn't to say block isn't important to consider in a general sense, but that it has no bearing or effect in an avoidance DR comparison in this context typically.

Remember, it isn't a comparison of how good defense is as a whole, but how each avoidance source is affected by DR.