Printable View
I've posted a question on the official forums about this, which you can find here, but I'm linking it here for our US numbercrunching friends as well.
I have yet to find any WotLK evidence of either case, so I'm currently clueless as to which way it should be. 415 baseline defense sounds the most logical (easier to program & can be ported to PvP with e.g. 399 defense skills etc.), but I can niether prove that theory, as I can disprove the current 0.6% theory.
Any thoughts/suggestions/help? :)
I don't understand what you're trying to get at.
Are you trying to figure out the specific avoidance penalty per level?
As in, "When a mob is x levels above me, reduce my avoidance by x*c, where c is the penalty per level."
I am not sure I understand what you are trying to figure out.
Also, the whole "no one questioned the -0.2% per level" is actually wrong. People spent all kinds of time testing that back in TBC and some were rumored to have before TBC. I know I personally spent a ton of time with Prince parses (yay for dual wielding for data collection) and adjusting my avoidance levels. -0.6 was spot on.
What changes do you think diminishing returns have on that? Those values are pre rating and pre rating values are not affected by DR (unless you are asserting those pre rating values are indeed the only ones that are).
What are you trying to figure out, exactly?
PS: This isn't a knock at you. I just am trying to figure out what it is you disagree with exactly and how DR fits into that. It wasn't very clear from either post.
The paper doll numbers assume your target has a weapon skill equal to 5*level
The level difference was/is worth exactly 15 weapon skill.
Up untill tBC 15 defense was spot on 0,6% avoidances.
Defense bonuses to avoidances are 0 if you have a defense skill equal to the attackers weapon skill.
Because defense was always equal to a certain amount of defense, no test in tBC could see the difference between adding a -0,6% avoidance penalty or a -15 defense penalty.
In other words, you could use 2 formulas for the defense gain on an avoidance:
0.04% * (defense - 400) - 0.2% * (your level - attacker level)
OR
0.04% * (defense - 5 * attacker weapon skill)
Both formula's were equal in tBC. But now with diminishing returns, the 0.04% * (...) part is now subject to diminshing returns, while the bold part in the first formula is not.
But I have yet to see any WotLK numbers for either formula.
That's really what I want to know.
I am pretty sure bosses are not subject to diminishing returns. The loss of avoidance isn't calculated as a loss from you. It is calculated as a gain for the boss. Plus, the 0.04% is subject to DR when talking about our gain, but the loss (or penalty) doesn't come from defense), it comes from the bosses weapon skill which is not subject to DR as of yet.
Think of it like this: Is our expertise skill (which is the replacement for our weapon skill on gear) subject to diminishing returns? If so, then the penalty would be. If not, the penalty would not. The bosses ability to negate your avoidance isn't subject to your avoidance DR.
That makes sense yes, sorta.
I'm aware that Expertise does have Diminishing Returns, but on the other hand, it does not negate Block chance or misses like Weapon skill, nor does it reduce Dodge/Parry in the same fractions as Weapon Skill.
But nothing of that really matters unless there is actual evidence to back it up. And so far I've yet to see evidence to support either (in WotLK).
Perhaps the clearest way to know is to check combat stats for % of misses, as those have the biggest diminishing returns. E.g. if someone has an Anub adds tanking log without defense rating procs and can tell us his exact defense for that fight, then it should give us some actual data to work with.
It seems like you're thinking about this backwards. It's not a penalty at all, for anyone, and diminishing returns has nothing to do with it.
Diminishing returns has to do with avoidance skill you get from avoidance rating. It has absolutely nothing to do with effective avoidance you get from your baseline stats, and that's what's at play here.
It works out like this:
A boss has 415 weapon skill. A level 80 player has 400 base defense skill. Because the boss has extra weapon skill compared to the player's base defense skill, he reduces the player's chance to dodge/parry/miss/block by .6% each. (he also gains .6% chance to crit as well, which is why the crit minimum needed is 5.6%)
Now, when the player increases his defense skill, he's gaining avoidance that is subject to diminishing returns, but that's AFTER the boss gets the bonus to the combat table from the level difference. Diminishing returns doesn't matter in the least.
The reason It's said players have a .6% reduction per stat against a boss is because the paperdoll displays your avoidance against a level 80 character, so you must adjust for a boss. In other words, the avoidance "penalty" isn't a penalty at all, it's just accounting for the displayed avoidance vs. a level 80 character and adjusting for the initial combat table against a boss.
To summarize: boss stats aren't subject to diminishing returns because they're not getting their stats from rating on gear, they are the base stats that they get from their level. That's where the avoidance "penalty" comes from, it happens before diminishing returns do anything at all.
Let's try this again:
There is an avoidance correction on bosses.
This correction has been tested in tBC (and maybe vanilla, not sure) and was proven at that time to be -0.6% for each avoidance. At that point 0.6% of each avoidance was worth 15 defense and the other way around.
At the same time it was also tested that weapon skill of equal value countered defense, easily proven by the old weapon skill racials.
This is all from years ago and has been tested tons of times, so I think we're agreeing on that being correct, right?
Now you're telling me that in WotLK, avoidance is first calculated, and then has this correction added to it SEPARATE from the part where the diminishing returns are calculated upon. I'm simply asking why this correction can't be part of the diminishing returns. Do you have any proof that it's still exactly 0.6% avoidance reduction vs a boss? Or is it possible that the 'base' defense after which diminishing returns start to kick in is actually 415, the boss' weapon skill level rather than 400 = 5 * your level?
The problem I have is that while everyone keeps hammering on that 0.6% avoidances, I've yet to see any actual data on it. People don't even know if this number is based purely on the level difference or the weapon skill vs defense difference!! I don't even know if defense reductions below 400 (thus negative defense gain from rating/buffs, such as Thorim) are linear or following the DR formula as well!! Maybe that's not something most tanksthink about, but it's exactly those numbers outside our regular scope that can provide us an extra insight into the formulas we DO rely on most of the time.
The biggest importance here would ofc be the 'unhittable' sets for anub, although there could be other fights in the future where such a difference could prove to be important.
Edit: example
Total Dodge chance subject to DR is 20.6%
DR formula applied gives 18.11%, minus 0.6% gives total of 17.51%
In the other possiblity, Dodge subject to DR is 20.6% - 0.6% = 20%
DR formula applied gives 17.66%, an increase of 0.15%
Perhaps not a world-changing number, but imo more than big enough to matter.
PS: It's safe to assume crit reduction does not have DR, or we would've noticed it by now. Perhaps there is irony in the fact that while one of the two 'avoidance' stats that can only be gained through defense has no DR, while the other (miss chance) has the steepest DR of all.
I think I can settle this, though obviously not with 100% certainty since we've not actually decoded WoW.
The game calculates a base comparison between attacker and defender (regardless of which is computer or player): Weapon Skill vs Defense Skill.
For players we can no longer change our weapon skill, but as tanks we increase our defense skill through defense rating (until Clysm anyway). The computer only uses set Weapon and Defense skills based on level. This is the baseline mind you.
So, since you're talking about the computer attacking you, the comparison would be your defense vs the boss' weapon skill (415 for a Raid boss/lvl 83). The comparison is the baseline, so your avoidance from defense is calculated on the remainder.
If you have 540 defense skill, then the defense difference is 125 and you go from there with calculations.
Looking it in terms of an avoidance "penalty" is slightly misleading, as there is no reduction or scalar, it's just a shift in the baseline difference. So, you can think of it as a reduction before diminishing returns, it is always that constant 0.2% per 5 pts difference to miss/dodge/parry/block granted by defense skill (note: not rating).
If you're concerned with reaching that 'unhittable' mark, it is sufficient to use the macros and get your total miss+dodge+parry+block chance to 100% + 0.8%/lvl difference (101.6% for heroic Anub adds).
Crit reduction does not face diminishing returns, and miss, logically, has the lowest cap as it is constantly available (meaning positioning, casting, etc will not remove your chance to be missed). I'm not sure why you think Dodge and Parry from defense don't have diminishing returns, Blizz said, point blank in WotLK beta period, that they do.
The standard disclaimer is that a great deal of what we do with this game mechanic formula work is things that we've either been granted a glimpse at, or deciphered. Blizz intentionally, rarely shows or tells us what is actually going on. Because of that there is always room for error on our end. The other side of it is that the game is written by programmers to be used in an online application. It is in the programmers interest to keep things simple and straight forward to make the game most easily playable and run smoothly. Because of that, much of our models are influenced by people savvy to programming so they can tell what is an unnecessary abstraction or an unlikely method of coding.
A bigger issue here is, whether or not it is dead-on 100% accurate, if there's a small factor of divergence from the real code, will you ever actually see the difference? These formulae are working for people applying the theory right now. Do you have good cause to question it? Data? Evidence?
I'll start with the origin of my question:
While building up the formulas for my spreadsheet I'm taking a relatively scientific approach. I am verifying all formulas I'm implementing for 3.2 accuracy, just to make sure I won't have to do twice the work and change them later on.
To explain the no DR on Dodge/Parry, I was pointing towards the Defense below 400, or baseline, as you refer to it, but I'll leave that out for now.
The thing you need to explain though is;
At first, you're saying that 415 is the baseline 0-gain level. All defense beyond that point is used for the gain to avoidances. So for a tank with 540 defense you get 125 defense or 5% dodge, which is then diminished.
But two paragraphs later, you seem to insinuate that rather than using a 415 baseline, you should use a 400 baseline (the paper doll one), and only once diminishing returns have been calculated, add the level difference.
Or to use the names used here, where do you account for the 0.6% avoidance from the level difference, on A or on Ad ?
Oh, no sorry, you misunderstand.
Your character sheet displays stats comparing your defense skill to a weapon skill of equal level. So if you're going by those numbers, that's where you remove the 0.2% from each stat influenced by defense per level of difference.
The game calculates based on the difference. 5% is the base value (not affected by diminishing returns) for each stat, so when your defense skill equals the attackers weapon skill, that is the base values (that still aren't affected by diminishing returns). If the target is 3 levels higher than you, it is your defense skill vs the new weapon skill (415), so you can erode that 5% of non-diminishable value by 0.2% per level difference, or 0.6% per stat then. All of this is before considering defense rating. Add defense rating inflates your defense skill and starts building you back up. If you had sufficient defense skill to bump you to 415 defense you'd just about counter that loss and be break-even at 5% but the earning back from defense rating is subject to diminishing returns for miss, dodge, and parry, so even at the very beginning of the curve it won't quite be parity (1:1).
Short answer to your question, the 0.2% reduction to miss, dodge, parry, and block chance from level difference is neither subject to, nor affects diminishing returns on your avoidance values.
And in terms of A or Ad, neither. A is your avoidance from ratings before diminishing, Ad is the avoidance after diminishing. This affects your "naked" stats, the stuff that doesn't enter into the diminishing equation.
So in other words, against a boss you say I should get:
5% + talent + DR(avoidance from gear) - 0.6%
with DR(x) being the diminishing returns formula applied to x
I understand that all fine, but what you still haven't explained is;
How do you KNOW, that that 0.6% is not subject to diminishing returns as well?
Is there any WotLK test data I was unable to find proving this 0.6% is not part of the diminishing returns? Because so far I've only been able to get everyone to explain to me how the formula works, but not why that formula is the correct one and others aren't.
If I play the role of an extreme sceptic I could simply state here that the formula should be;
5% + talent + DR(avoidance from gear - 0.6%)
because defense rating on gear gives defense, which is added to the 400 innate from level you (should) have and after substracting the attackers weapon skill, you get the value which you need to multiply with 0.04% and add to the avoidance you get from other ratings, and apply diminishing returns on that.
There ya go, that's a simple formula, with a somewhat plausible explanation. Now you're telling me that I'm wrong and I could simply quote you backYou say the formulae are working, but in the spirit of scientific scepticism I could say; Why should I take your word for it? What testing has been done to find out if this level difference thingy is added on the DR or not? What speaks in favor of the first formula other than the fact that more (famous) people think it is correct?Quote:
Do you have good cause to question it? Data? Evidence?
An example:
With 700 Defense (something close to your average unhittable set I presume), you tell me that the chance to be missed by a boss is 18.92%. I just want to know why this can't be 19.48%, adding the 0.6% into the DR formula. I'm not familiar with any online combat log parser that also shows the tanks defense at that time, nor do I trust Armory to show people in the gear setup they used, so I can't really find any evidence unless someone goes out, gets swung at tons of times & says: Here is my data, here is my defense, see what formula is the best approximation and stop the discussion already!
No, I haven't done, or seen testing specifically from WotLK to show that is still the case.
The understanding was developed during BC, yes, and tested to prove it's consistency, which is what gave cause to the understanding that it is a root change, a baseline change in the formula. There has been no reason to question that changing in WotLK because of that concept of how it works.
And, I'd adjust your formula as follows:
5% (base) - (0.2% * attacker level) + talent% + (diminished avoidance from ratings)
So, there are two points we can go from here (other than trolling EJ or other sites to see if anyone has actually tested it):
Do you have reason to believe that this is not the way it works? Do you have reason to not accept the concept that this is a baseline application to defense skill vs weapon skill (which are not subject to diminishing returns)? If you don't think it's coded that way, that's fine, we just have to find a way to test it, but there's nothing that has changed from BC to WotLK that we have any indication, evidence, or bulletin about that would reasonably change that, so this would be working to find a stealth change, or a change to something we don't/didn't understand.
At that, is it wildly important that you figure out if it is the case? Where is it applied currently? Heroic Anub add tanking "unhittable" tank gear sets? If the 0.2% is subject to diminishing returns where is it in the math/coding/game mechanics? If it is subject to diminishing returns, how much will they change that value?
Most of our game theory is built on suppositions, we don't "know" that it is the actual case, but it is sufficient to say it is a best guess until tested and disproven. Please do post up if you find someone who has tested it, or if you do some testing yourself. I'd be interested to know if the concept I've been using for years is not actually the case any more.
Because neither of the two values it comes from (415 weapon skill and 400 base defense) are subject to DR. If they themselves are not subject to DR, how can their difference be subject to DR? It doesn't make any sense.Quote:
Originally Posted by WarTotem
But how do you know you need to use 400 defense as base vs 415 weapon skill?Quote:
Originally Posted by jere
And if that doesn't make sense;
Naked Agility is not affected by diminishing returns, but agi from gear is.
And that's not even a universal base, every race has a different base there!!
Quite often formulas grow over the years and get adapted as the mindset of the developers change.
Before WotLK, you could calculate avoidance from defense using (your defense - attacker's weapon skill) * 0.04% and it would always be accurate even in PvP.
This formula would require less calculations than (your defense - 5*your level) * 0.04% - (attacker level - your level) * 0.2% especially since you need to add in your level twice.
Why did you add the redundant factor in the last equation?
And the question remains, why do you think this doesn't still apply, or are you just looking to question it?
What we're saying is, nothing they've changed that we know about affects the concept we've had from before of how this works.
If you're looking for something more solid than that, it looks like you'll have to find a way to test it.
It was tested back in the day. Nothing concerning DR would have changed that though (since DR doesn't apply to either of the base values), so they would have had to manually change it specifically outside of the DR portion. What makes you think they did?Quote:
Originally Posted by WarTotem
Your agility comparison is a totally different comparison than saying that X and Y are not affected by DR so (X-Y) is also not affected by DR.Quote:
Originally Posted by WarTotem
Your naked defense and your naked dodge are not affected by DR either while defense and dodge from gear are. Nothing inconsistent about that.
The equation you listed would work before WotLK for sure, but that doesn't really change anything. The base values were tested by Whitetooth after WotLK came out and he found that they didn't get affected by DR (and neither did naked values). Now if you think his findings were faulty, then he might still have the data posted up at EJ you could sift through to help you find support for that (though he did clean up the thread a couple of times, so not sure). A lot of us kept up with those results at the time. They looked sound based on the data he collected. I would find it odd though if he were wrong, as the DR equation has been solidly working since it was formalized.
These are the knowns:
1. The -0.6% was determined back in Classic wow as the difference between base weapon skill and base defense skill for PvE (don't know PvP and PvP is often different). If that has changed, they would have specifically had to do it because nothing they have announced (DR or otherwise) has changed that affects base values. Could it have changed? Maybe, but not likely. Still, there is no harm in testing it if you want.
2. Research was done after WotLK came out that determined that both base and naked values were unaffected by DR. Again, the 0.6% was from base values. It doesn't matter if an earlier equation which had base values in it worked and no longer works now. That equation also had values from gear (which were changed).
I guess the best you can do is test it and see, but really, there is no visible reason to question it at this time. Still doesn't hurt to question it, but nothing we have brought up sheds any doubt on the existing mechanics yet.
If you do want to test it, you need to gear up in a bunch of pvp gear (for high stamina and resil) and have someone heal you as you solo tank a raid boss multiple times (you will need a whole lot of data to verify a delta as small as 0.6%...a whole lot of data). You might be totally right. But really only lots of data will support or disprove it either way.
As a note, I am not saying you are wrong to question it. There just isn't anything so far that would indicate that it could have been changed. So I am just skeptical about being skeptical. Normally, there would need to be some sort of reason to question something.
Still, testing for the sake of testing and re-verifying is still useful stuff. So forgive me if I sounded too negative about it.
Can you say what exactly you mean with redundant? Because I'm not sure what part you're pointing at.Quote:
Originally Posted by Satorri
And you could say I'm questioning it simply because I want to, but on the other hand, I am used to using 0.04% * (defense - attackers weapon skill) in the old tBC formula instead of paper doll - 0.2% * level difference, because both gave the exact same result.
But I'll do some tests later this week if I can find a suitable testing environment.
@jere:
Compared to tBC, diminishing returns has changed.
I'm assuming you're referring to this thread on EJ?
I tried skimming through the tons of pages of test results there, but I couldn't find any results indicating if the level difference adjustment has to be done on the final number or the number subject to diminishing returns.
Oh, and as test case I'm actually aiming for those giant skeleton guys in Icecrown, they're lvl 82 iirc, which shoul still prove a substantial difference either way on misses if I max out my defense.
Yep, those skeles might be good. You just have to make sure not to kill them by accident! I don't recall them having too much HP's