Olgas

Please read the thread you are posting in before making a useless post like this.

The testing described in this thread has proven that both of the things you are assuming to be correct, are, in fact, incorrect in several important respects.

 

Olgas

Quick and dirty summary of where we are so far:

1. Your base miss rate against any mob or boss is:
And, of course, if you are dual wielding substitute 24% for 5% in the above formula.


2. If the difference between the Mob's Defense Skill and Your Weapon Skill is 10 or less, then X = .1%.


3. If the difference between the Mob's Defense Skill and Your Weapon Skill is greater than 10, then we think X might be .266667.
(This is what we are in the midst of testing. We seem to have disproven some other theories on what X might be).


4. In doing the (Defense - Weapon Skill) calculation, fractional Weapon Skill amounts are dropped. For example, if you equip gear that adds +15 Weapon Skill rating this would add 3.804 to your Weapon Skill. However, it is treated as 3.00 Weapon Skill for purposes of these calculations.


5. Applying the above formula yields the following base miss rates for a Level 70 character with 350 Weapon Skill:
Note that we are still confirming the 9.00% figure. But we know for sure it is very very close to 9.00% and currently believe it is exactly 9.00%. (Also, obviously, just add 19% to all the figures above if you are dual-wielding).


6. Because the formula is dependent upon the differential between Weapon Skill and Defense Skill, adding Weapon Skill (whether through talents, racial bonuses or gear) will reduce your miss rates against all mobs. But, because of the way the formula works, the beneficial impact of adding Weapon Skill is non-linear. In particular, as a Level 70 character, adding the first 5 levels of Weapon Skill (i.e., getting your Weapon Skill to 355) is the most beneficial, because it will effectively reduce your base miss rate against Level 73 mobs/bosses from 9% to 6%.

 

0nighthawk0

Thanks Olgas,

I'd made quite a few spreadsheet of my own with my own testing, and just had to confirm that for 71's 72's etc...1 point of weapon skill (not rating) actually appeared to add 0.1% hit as well as the supposed 0.1% crit stated in patchnotes for mobs above 70 (in fact more data to prove hit that the crit lol) ....as i ran accross this quite accidentally during calculations when tightening things up. So i signed up here having read a few posts pointing in this direction, because vs level 73 mobs and bosses my calculations appeared to be off (not working in an entirely linear fashion comparitive to Lvl71,72 etc). The 72->73 gap difference is the subject of much debate and quite a lot of misinformation on the net (especially with blizzards changes over the WoW lifetime confusing matters).

So thanks for clearing that up and showing the state of play so far with findings and testing, i understand a lot more about whats going on and where to modify calculations made.

Keep up the good work

 

Emeraude

Olgas, thanks much.

 

Herb

I'm wondering why you're assuming the formula above for skill differences >10: didn't the wowwiki formulation itself introduce the concept that base miss and skill difference aren't universially linearly related, in spite of what differences <= 10 might have implied?

Since it's original incarnation (with X = 0.2) has been refuted, why stick to it's general, linear form as a working theory? The continuous alternative (Spoon/Karmon) has been disproven, but these two weren't the only alternatives to start with, of course - for all we know, we might just as well have to to deal with a table-based function.

 

sp00n

If it was a table-based function, you'd require entries for every kind of defense to weapon skill ratio above 15 points difference as well. I.e. not maxed out weapon skill, or a level 69 player fighting a level 73 mob, or the above mentioned mind controlled protection warrior with 511 defense.

And a linear function -> table-based function -> linear function switch to get rid of this problem seems to be quite unlikely.


As for the current theory, maybe it's even a 2% jump when the difference is >10.
I've tried to come up with another formula that proposes a 1% jump every 5 (6) points of weapon skill (defense), but that was ruled out by the ~2000 swings without a miss with 316 hit rating here (would have been an expected miss rate of 1%).

And with this 2% formula, we could actually make use of Crezax' posting (0.8% with 4 weapon skill):

So you'd need (24+4-5-(2*0.2))*(10*82/52)= 356.39 hit rating to never miss with 352 weapon skill.

 

Olgas

Well, I'm still trying not to "assume" anything at this point. We're still postulating and testing. And I was even a little reluctant to write that last post because there is always the risk that you confuse people by giving them an update before all the testing is completed.

The truth is, while we've learned quite a bit over the last several weeks, we still don't know the exact formula for skill differences between 10 and 15. We are pretty darn sure of the miss rate at a skill difference of 10 and at a skill difference of 15. But we're not sure what happens in between, at skill level differences of 11, 12, 13 and 14. The "modified Wowwiki" formula with .26667 is one possibility. The new formula Sp00n just entered above is another possibility. And your table idea might also apply.

 

Tabrab

So yesterday I was able to get exactly 89hit rating with 359 weapon skill with a 2h axe. We went on to Nightbane and started logging. And here what happened:

8/18 23:59:32.012 Your Execute missed Restless Skeleton.
8/18 23:59:36.904 Your Hamstring missed Restless Skeleton.

So with the post #300, it says that I should never see a miss... Granted it wasn't a miss on Nightbane, but on a lower level mob...

edit:
ok, never mind, just missed this line:
8/18 23:59:31.774 You are afflicted by Distracting Ash.

 

Olgas

In posts #306 through #312, we discussed a test Arcazua did against a mob FOUR levels higher (20 level skill differential).

He had 178 misses in 2000 shots (8.90%). 95% confidence range = 7.65% - 10.15%.

Against a mob +4 levels the working theories would yield the following miss rates:

Sp00n/Karmon: 12.00%
Modified Wowwiki: 10.33%
New Sp00n idea: 10.00%

Interesting.

Let me try to formulate the "New Sp00n idea":

It's actually a very interesting idea. It has the virtue of being consistent with all of our tests to-date (I believe), including Arcazua's +4 mob test. And it also has the virtue of being consistent with Crezax's original post...

 

Tabrab

Now I feel like an idiot... First time posting, just woke up, didn't double check my log file... Though I will continue to monitor this current setup I have closely on our next boss fights.

 

Olgas

Phew!

I was scrambling on Bosskillers to see what debuffs could possible explain the miss, LOL.

Thanks for posting the correction. :-)

 

Eternity

I don't know if this might help you.

I had 1 miss on morogrim on a long ssc run with 297 hit rating and 316 skill rating sword.
you can check how many hits on moro lurker ect...

wws

 

sp00n

297 hit rating is 18.83% to hit, +5% through talents is 23.83%. 316 weapon skill (not rating) would mean a 23.9% chance to miss a target, so perfectly in line with the current <10 difference theory.

 

Herb

That is exactly what I hoped for - we can't assume overly much yet for that critical range difference; the "table idea" was merely there to illustrate that (since that would allow arbitrary base miss values for each difference).

 

Giske

So am I correct in assuming that I need 298 hit rating with 366 weapon skill to reach the cap versus a level 73 mob with this new formula?

 

sp00n

298/(10*82/52)= 18.89, so no, you'd need 299 hit rating with 366 skill: 299/(10*82/52)= 18.96.

 

Giske

Right, I was getting the same thing, I just wasnt sure if I needed to bump it to 299 because of fractals or not. Now I have a goal, but how I'm going to reach it without stacking hit ridiculously I dont know :p

 

Eternity

It's more likeky to be 298, isn't it ?

298/(10*82/52)= 18.897

isn't it 366.34 skill with vashj belt ? so 23.866 chance to miss.
18.897 + 5 = 23.897


i got 1 miss out of 3000+ melee swings

297/(10*82/52)= 18.834 + 5 = 23.834 for 23.866 miss chance
meaning a 0.0318 chance to miss, isn't 3/10000 or 1 for 3333 ?


(my math is pretty bad, sorry if i'm wrong)

 

songster

Weapon skill ignores everything after the decimal, so it's 366 skill.

 

Darkrenown

Hmm, so does this mean that with a 2h and 359 skill I only need 92 hit rating?

 

sp00n

No, with 359 skill you're facing a 5.6% miss chance against a level 73 mob, therefore only requiring 5.6*(10*82/52)= 88.31 = 89 hit rating.

This is, if 2h is under the same formula as 1h, which currently is not proven but I don't see a reason now why it shouldn't (just like ranged attacks seem to share the same formula).

 

rhea

So who thinks crezax's post was anything but his own speculation on how weapon skill works?
afaik he is only a community manager and not a developer or anyone with access to their formulaes.

Anyway, anyone pushed weapon skill way up?

 

Darkrenown

Even better. I'll try it out tonight then.

 

sp00n

Did you bother to read the thread?

 

Olgas

Sp00n, I like the testing approach you are taking, i.e., equipping enough +Hit gear to the point where the theory says we should not see misses, and then looking for misses.

It seems we still have at least two theories to test
-- the "modified Wowwiki" theory (5% + Difference*.26667%); and
-- the "New Sp00n" formula (6% + Difference*.2%)

I was trying to think how I might help test these -- but as a Hunter I'm really limited in my ability to equip ranged Weapon Skills between 351 and 354.

So, I was trying to think up some test cases for melee fighters. Here is what I came up with:

Proposed tests:

Dual-Wielding Rogue: Precision and 354 Weapon Skill (2 points in Mace Specialization?)
-- First test case with 346 Hit Rating
-- Second test case with 351 Hit Rating

and/or

2H Paladin or Warrior: 354 Weapon Skill (2 points in Weapon Mastery or Weapon Expertise?)
-- First test case with 126 Hit Rating
-- Second test case with 130 Hit Rating

If we see any misses with the first test cases, then we will know the "modified Wowwiki" formula is wrong.

If we see zero misses with the second test cases, then we will know the "New Sp00n" formula is likely correct.

This of course still assumes that the base miss rate against a Level 73 boss is exactly 9.00%/28.00%. I'm still running my tests to validate that.