vytautas

Took some 2H data

4 Expertise from spec = 1%
18.00% crit on character sheet - 182 agi + Cruelty + Axe Spec
Attack Speed 3.15 57 Haste Rating + Blood Frenzy
0 Hit Rating

Only items worn, tabard + Betrayer of Humanity



Obviously the sample size is small, particularly for the yellow attacks but is it possible the hit required for yellow is still 9% and 8% for white? Then again with the deviation I had in dodge vs expected it could all be meaningless, I'll try and bring myself to do the same thing tomorrow night and add to the sample.

 

• Aldriana

With a sample size of 2569 your 95% confidence interval for miss rate is 7.89% to 10.09%, which easily encompasses both 8 and 9% as options. So I don't think we can conclude too much from this; I think the far more effective way to verify the yellow miss rate would be to equip exactly 8% hit (or slightly more, but as close to 8 as possible) and perform a couple hundred attacks; if the miss rate is 8%, all will hit, but if it's 9% you should get a miss within the first couple hundred attacks.

 

Xerophyte

A 95% confidence interval for your data places places the actual miss percentage between 26.5% and 28.1%, which unfortunately doesn't really help distinguish between our probable options of 27% and 28%. Like Aldriana said the preferred method for testing for these scenarios is to gear up at just above 27%, then do a few hundred attacks and see if any hit. The sample size required to determine the DW miss rate by hitting dummies while naked is prohibitively large.

 

Revart

here is whats current of an on going test (letting it run while I'm work) I don't really know what this means butI'm sure we'll get to the bottom of this, also can any one explain that expertise?[EDIT]: I'm stupid its from a talent, also I did the math on the expertise and its about right anyways, I'm going home at lunch to check on it, I'm going to take what ever my % to miss is and try and put on that much hit and see if it clears misses from the table, so far as of now, I think my theory on haste is pretty much out the window.

 

vytautas

This is what I get for never paying any attention to maths ;< Knew there must be an easier way, will set about the below immeadiately if only to make myself seem slightly useful rather than just an idiot ;p Getting 27% hit might be harder but I'll work on it later.


- No Execute misses.

- No Melee misses.

 

◊ Sarutobi

EDIT: Results unreliable as I didn't have the dummy to myself. There may or may not have been other chars applying +hit debuffs to the dummy making any data useless.

 

• frmorrison

There aren't any +hit debuffs other than Misery/Imp FF, but those give spell hit.

The only one is Draeni Aura, but we don't have to worry about those.


After doing a Naxx 25 clear today, I had 0 misses with 5% hit.

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Exemplar

Take frmorrison's stats with a grain of salt. Some Paladin are finding they're at 5% hit cap, some 6, some 8. It varies paladin to paladin regardless of spec for no known reason. Don't take this as a global all-classes statistic, just that he personally hit caps at 5% (lucky sod).

 

Leart

I ran a small sample today. I'm not sure it's enough to qualify as anything for you, but here it is.

1007 naked with a white polearm auto attacks:
5 Expertise from talents
0 Hit
9.1% miss

That was in unholy presence (15% haste), with improved icy talons (4% haste).
I'm not certain haste is having any odd effect.

 

Autoband

8% would be sensible seeing as the taunt glyphs give 8% hit, and not 9 which would be logical if the cap was indeed 9%.

 

Multane

If i'm not mistaking Taunt uses the spell hitcap, which was believed to be 17%. I have no idea if this still holds true, but 8% from glyph would leave 9% to get from gear.

 

charriu

But on the other hand, you need less hit rating to get those 9% spell hit than you would need for 9% melee hit.

From Combat Ratings at level 80:

Rating	Level 60	Level 70	Level 80
Hit	10	15.76923275	32.78998947
Spell Hit	8	12.61538506	26.23199272

 

Autoband

Though normal hit already increased chance to hit taunt long before hit and spell hit were unified. If it does use the spell hit coefficient then 8% melee hit would be about 10% spell hit giving you more than the 17% needed with the glyph. This might just be badly thought through by blizzard though.

 

• frmorrison

Incorrect, Taunt has always used spell hit. Back at level 60, some Warriors used spell hit item when they were tanking 4 Horsemen, and one of their set bonuses gave a lot of hit to Taunt.

Since hit isn't that strong of a tank stat, it makes sense to give a lot of hit to Taunt, especially since it needs a lot of hit to make it land on a raid boss (which most are affected by Taunt).

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Atros

Right now, I'm starting to do some naked tests as a cat druid. I'll put up some recount screenies soon, but with a small sample size, it seems to be floating between 8 and 9%. Way to small to say anything definitive yet, though.

 

• frmorrison

The druids already decided their hit cap was 8%, and even updated Rawr with that info.

DK - Ashbane Failure is the condiment that gives success its flavor.

 

Atros

Right now, I'm under the impression the jury is still out, with the desire for more data. Rawr has been updated to show 8% as thats what the new data is leaning toward. That still doesn't mean that we can't say for certain what to expect.

 

hellord

I did some tests today against Boss level dummy in org.
Results were in favour of the 8% hit cap.

I'm arms 2h axe, 8.05% hit (262 hit rating) 15 epertise (50 rating, 3.75%) 33.20% crit (battle stance, 650 crit rating)

Total count (melee + specials): 1788
Total crits: 619 (34.62%)
Total dodge: 47 (2.63%)
Total miss: 0

White hits

Count: 720
Hits: 337 (46.81%)
Cirts: 219 (30.42%)
Dodge: 19 (2.64%)
Glance: 145 (20.14%)
Miss: 0

Specials

Count: 1068
Crits: 400 (37.45%)
Dodge: 28 (2.62%)
Miss: 0

Thunder Clap

Count: 40
Hits: 30 (75%)
Crits: 6 (15%)
Miss: 4 (10%)


Test seems correct since:
a) dodge values are inline with expected 6.25%-6.50% chances. Overpower can't be dodged but can be missed, and prolly heroic trow too (didnt test this anyway)
b) Miss value for thunderclap seems inline with 17% spellmiss (8% from gear, 9% expected, too few samples)
c) 0.95% chance to miss would have produced around 16 miss in totals.

The zip file contains an xls report and screenshots from recount (I wasnt logging, i'll do more tests btw)
http://www.hellord.com/wow/tests8_hit.zip

Here's the white damage screen with hit rating/%

ArP Whore

 

sephfiroth

Before Wrath was released, some Paladins including myself speculated that if you had 3/3 Precision talented(before they removed it) you still retained that 3% Hit. Not sure if that was ever fixed, or if a definite conclusion was reached.

 

Exemplar

Testing proved unrelated to Precision. Some Paladins who had Precision have 5% hit cap, some have 8%. Some Paladins who did NOT have Precision have 5% hit cap, some have 8%. It is not based on a current talent, previous talent, aura, etc. 0/0/0 specs who had before/after come up with same results (both 5 and 8). Needless to say it is confusing.

Paladins at "worst" match other classes with 8% melee hit cap. At "best" their hit cap is as low as 5%. This varies from paladin to paladin. Why are worst/best in quotes? Having 5% hit cap is debatable as a good thing, since it will inevitably be corrected (we hope sooner rather than later) and basing your gear around this would alter things greatly when it is removed. Not basing gear around this leaves you heavily over the hit cap, and therefore under your own best possible performance.

 

Trinsian

I'm getting numbers from both myself and my pet that are trending towards the hit cap being 8%. Wanted to verify for myself before I believed it, but as a Draenei I'm coming up on close to 10,000 auto shots naked (with a level 4 gun), and a similar number of pet attacks (started attacking with my pet after around 3000), and both sets of numbers are trending toward 7% (1% draenei hit bonus) for both myself and my pet. I haven't added up all of the tests yet so as a whole this is still anecdotal, but the test currently running is:

Me(No focused aim):
Hits: 2364 (92%)
Misses: 186 (7.2%)
Crits: 20 (.8%)

Total: 2570 (100%)


Pet(Has crit, expertise talents, etc):
Hit: 2109(46.5%)
Glancing: 1103(24.3%)
Crit: 831(18.3%)
Miss: 327(7.2%)
Dodge: 170(3.7%)

Total: 4540(100%)

I have several other segments I haven't added in yet, but recount shows all of those tests as being around 7% as well (some slightly under, some slightly over, all with over 1000 attacks each). I'm not doing WWS for all the stuff I've run today, but have SS's of the Recount boxes if anyone doubts the numbers. At some point in the future I'll add all my segments up, but this is a relatively large sample size by itself, and my total results from today are enough to convince me.

*Edit: Interestingly (although a subject for another post) my crit rate is almost a full 5% lower than that on my paper doll.

 

Larsiak

Seeing no really conclusive tests about dw-base miss being 27% or 28% against boss lvl, i decided to run a test myself.



I spent about 8 hours on the heroic training dummy in UC split in 2 sessions.

WWS Session 1
WWS Session 2

I was completely naked except 2x [Poniard] due to weapon speed, and paid attention to their durability not droppping to zero.
I had maxed-out dagger skill, 0/3 precision and accurately 4 expertise from Strength of Arms

I accidently hit Bladestorm once, but it's only a cosmetical issue in recount and has no further impact on the data. (I have no clue why it doesn't show up in the WWS.)
In session 2, I was in zerker-stance for a few seconds but i didn't got any crits during these few hits.


I decided to calculate the percentage values with two relevant decimal places. Feel free to calculate with more while checking my outcomes, but personally I see no point in it with this sample size.
I chose a 99% confidence interval for personal preference.

Sample size = 45.901

	count	arithm. mean	std. deviation	99% conf. interval	w/o 4 expertise
miss glance hit parry dodge crit block	12.290 11.070 9.256 6.037 2.548 2.395 2.305	26,78% 24,12% 20,17% 13,15% 5,55% 5,22% 5,02%	±0,21% ±0,20% ±0,19% ±0,16% ±0,11% ±0,10% ±0,10%	[26,15%;27,41%] [23,52%;24,72%] [19,60%;20,74%] [12,67%;13,63%] [5,22%;5,88%] [4,92%;5,52%] [4,72%;5,32%]	[26,15%;27,41%] [23,52%;24,72%] [17,60%;18,74%] [13,67%;14,63%] [6,22%;6,88%] [4,92%;5,52%] [4,72%;5,32%]

I feel quite safe to say 27% is the right value.
My outcomes back up about 24% glance-chance and about 4,6% crit-suppression.
Furthermore they suggest 26/56 as expertise soft/hard-cap.


Assumptions made:
- there are no bugs related to the training dummys
- white hits use a one-roll combat table
- expertise behaves like the tooltip says (eg. 4 expertise lower your chance to be dodged/parried by 1%)

All these are general consensus on these boards, so i have no bad feelings about them.


Finally: Go on and point out my silly mistakes.

ps: To nail down the first decimal place of the crit-suppression, you need a std. deviation way below 0.1% and therefore a sample size around one million. (3x std. deviation = 99%c onf. intervall < 0.06% -> std. deviaton < 0.02% -> (0.1%/0.02%)² * 40k = 10^6)
I doubt that anyone has enough spare time .

 

• frmorrison

That is intended. Physical dps have a 4.8% crit reduction when attacking a boss mob. Casters also have a similar mechanic, but it is not known to me what that number is.

DK - Ashbane Failure is the condiment that gives success its flavor.

 

RareBeast

It seems to me that a lot of these tests would be far more useful if they weren't done naked. A 10,000 hit sample with you sitting at 8.01% hit showing zero misses would be pretty hard to argue with. Same with the TG & 2-hand miss %'s. Set your gear up so you are just slightly over the lower suspected +hit % then perform the test.

Testing naked just requires an absolutely mammoth sample size to get any sort of reasonable confidence interval.

 

sp00n

Melee Hit to affect Taunt was introduced in Patch 2.3.0.
So Autoband is correct, as melee and spell hit haven't been merged together before patch 3.x.



Well, yellow attacks suffer an inherent (additional?) crit reduction, due to being on a two roll system. So only landed (hitted...) swings can crit.
However, adjusting your sample to a two roll system doesn't provide more significant results. Your relative crit chance over all your landed swings was 14.14%, and your expected absolute crit chance would have been 11.23%.

So about the same amount difference, just in the other direction.

I think Aldriana has already figured out that special attacks do also suffer a 4.8% melee reduction. At least that's what I just read in the other thread.