Is melee hit cap really 9%? - Page 5

Fwip · 12/31/08, 8:14 PM

So, I set myself up with 8.02% hit (263 rating) tonight and afk'ed behind a target dummy for 1600 swings. Out of these, I found 0 misses, 76 dodges, 100 crits, 400 glancings, and 1024 normal hits.

With 8.02% hit and a 9% base miss chance, the odds of me seeing 0 misses in a sample size of 1600 is: (0.9902^1600) = 1.4344e-7, or 0.0000143%. Seems pretty obvious that the miss chance is no longer 9%.

If we do the same math assuming different base miss rates:

Miss        ErrorProb     Odds that the base miss chance is lower than this
9.00%       1.4344e-7     99.999985%
8.50%       4.535e-4      99.954%
8.25%       0.02511       97.489%
8.10%       0.27789       72.21%

Screenshot

I also have 400 devastates logged without a miss, giving a confidence of 98.05% for the 9% miss scenario. It's not nearly enough to confirm that yellow misses are below 8.5%, but it is almost certainly not 9% anymore.

Edit: Swung a few more times, up to 2751, still no misses. Unless raid bosses are different than the heroic training dummy (which may be the case), the white hit cap is almost certainly no more than 8.02%.

tclan · 01/02/09, 3:04 PM

Long (some, actually) time lurker, first time poster.

I've been following the discussion here and I decided to do a test on the heroic training dummy in Orgrimmar.

Relevant stats: 263 hit rating (or increases your spell / melee chance to hit by 10.03% / 8.02%) + 3% spell hit from the death knight talent Virulence.
Attack speed after haste and such effects: 1.37 (was using [Ethereum Phase-Spear], the fastest two-handed weapon in game that I could find).
Weapon skill was maxed at 400, I was not grouped with anyone and I had the dummy to myself for most of the time and I am a troll.

WWS report

No misses in 6000 melee swings. No misses in 3036 special attacks (Heart Strike + Death Strike). 40/761 = ~5.3% miss rate for Death Coil, which is a bit high considering the spell hit cap is 17% if memory serves right. But I'm sure that can be explained by the fact that the amount of death coils I did is still a bit low to be statistically reliable. What kind of surprised me was that 22 out of 550 (4.0%) Dark Commands (taunts) missed the dummy. I had always thought that taunt used melee hit cap rather than spell hit cap, and that would make sense with the +8% hit glyph as well.

Well, just dumping my data here, someone else can make heads and tails out of it.

PSGarak · 01/03/09, 8:34 PM

Taunt (and suchlike) tend to be spells and use the spell hit cap. They were retconned some time ago to be affected by melee hit rating. The 8% hit glyph actually makes the most sense as covering the gap between the two hitcaps.

By the way, that gives rise to a question of my own. Now that ratings are unified, melee classes will end up with a certain amount of spell hit on their gear as a matter of course. This would seem to make change to melee hit rating redundant for taunt. Since your Dark Command was still missing it's safe to assume it wasn't double-dipping. But are they, in general, still using the melee rating->% converstion rate or have they been changed back to the spell rate? Note that spell is preferable, you gain more percent per rating.

Prinsesa · 01/06/09, 4:29 AM

If the spell hit cap is indeed 17%, then a 4% miss rate for Dark Command would be right on the money if it was using the spell hit conversion rate, since 10% spell hit from gear + 3% spell hit from Virulence would indeed reduce 17% miss to just 4%.

If it was using the melee hit conversion rate, then we would expect a 6% miss rate.

Of course, this would still be subject to confidence, since he only cast 550 taunts.

Cymraes · 01/09/09, 12:24 AM

Originally Posted by Origence

Wow Web Stats
On that wws I had 13,71% hit and 6.25% dodge reduction.
The dodge part worked perfectly as I had 0,1%-0,3% dodged attacks on various bosses which confirms the 6.5% cap for dodged attacks, nothing new here.
When I checked the report I saw there was a few misses on special attacks so I assumed everything was working alright and at 13,71% I could still miss. Then reading this thread and other warriors saying they dont miss with less I went to check the wws again and noticed that the only special attacks I missed were at Archavon the Stone Watcher. Also if you looks swings miss rate it was susbtantially higher against Archavon than all the other Naxxramas bosses. Which is surprising too.

A big part of the reason that your misses against Archavon were higher was probably because of his ability "Choking Cloud" It gives you a 50% chance to miss on attacks.

brutalbovine · 01/09/09, 1:18 AM

It looks like we have official confirmation on Paladins and Priests retaining +hit% bonuses from old talents after those talents were removed. From World of Warcraft - English (NA) Forums -> Hotfix: Talent Issue

During the transition from patch 2.4.3 to 3.0.3 the following talents were not unlearned properly by some players, resulting in those players still gaining the benefits despite the talents no longer existing in the talent trees. As a result, we've pushed a hotfix which has corrected this matter and affected players will longer receive credit for having these talents.

Warlock: Improved Lash of Pain
Warlock: Fel Stamina
Shaman: Totemic Mastery
Shaman: Improved Lightning Shield (Rank 3)
Paladin: Precision
Rogue: Fist Weapon Specialization
Priest: Shadow Focus (Ranks 4 and 5)

Apologies if this was already posted.

ksearo · 01/22/09, 10:09 PM

We all need to take a step back here and remember how the math behind what we are talking about works. If you had enough hit to put you 1% below the hit cap, whatever that number may be, you could still have more then a 1% of your total attacks miss. If you had a 99% chance to hit, it is still in the realm of possibilities to do 100 attacks (or even a million attacks) and never land a single hit. A 99% chance to hit is also a 1% chance to miss. The chance of getting one miss is 1%^1. The chance of getting two misses in a row is 1%^2 (0.01%). The chance of getting 20 misses in a row is 1.0E-38% chance. Of course this is extremely unlikely, but hopefully you are getting my point.

Multiple tests across all classes while using a 2 hander (no TG) is the best way I see to hone in on what the hit cap is. The larger the data pool the closer we are to the true value.