Thayer

I was doing some testing hunter side and I came the the same conclusion. Originally I thought it was that my crit from talents was not applying, but I eliminated that.

Here is a partial re-post with some relevant information.

Well, I am testing without lethal shots right now, and at over 2000 shots, I am still getting about 5 percent less crit then my tool tip says I should. I am testing auto shot only at this time. I did notice the fluctuation in crit and Hit on different abilities, as well as a variation in my pets crit, hit abilities. If the reduced crit is not tied into talents, as I seem to be finding, then can it be that they changed the crit reduction between levels to be a lot more drastic.

Once I get some 3000 shot test done, I will post screen shots of my returns, and we will see if anyone can duplicate them...

I ended up doing a 4000 shot test, since i saw some RNG fluctuations around 2000 shots. Here is a link.

http://i179.photobucket.com/albums/w...NoLethal83.jpg


In this test, my tool tip said 23.45, but I only crit 18.7 percent, so that is a 4.75 percent crit reduction versus a level 83 boss dummy. This is without Lethal Shots or Focused Aim, as seen in the screen shot, and i will also do test with Lethal Shots and no Focused Aimed. This will be followed by a test with Lethal Shots, and Focused Aimed, with 5 percent hit from gear.



Next I will do a 2000 shot test on a level 80 dummy, to confirm equivalent mob level crit rates.

I did the level 80 dummy test and my crit returns were in line with the tool tip:

http://i179.photobucket.com/albums/w...NoLethal80.jpg



It seems that there is a larger crit penalty for the difference between player / target level differences.


Since as a hunter I don't have to calculate for dodge and glancing blows, you can see that these are not effecting crit returns.


here is the thread link, so we can simultaneously keep track of new information.

Lethal shots not effecting my autoshot or steady shot - Elitist Jerks

 

sp00n

Very very nice testing Vulajin.
To recapitulate your findings, I think we can quite safely assume now that:

• Chance to miss is 27% while dual wielding (885.33 rating).
• Chance to be dodged is 6.5%.
• Crit reduction versus boss level mobs is 4.8%: 0.6% per level and an additional 3% due to a weapon skill to defense difference of >10.
• Glancing mechanics haven't changed.

// Edit
You didn't have any poisons applied during your testing I suppose? Could have been interesting to see resist values and crit reduction there too, but due to the difference in crit rating conversion and agility to crit, I suppose this would be inconclusive while focusing on melee mechanics.


PS:
I wish I had those test dummies available back when weapon skill was being tested.
And that 'somewhere' was in your thread, by me.

 

• Vulajin

Thanks, yes, I had trouble finding it because it wasn't in this thread and I have a billion other rogue threads subscribed. :x

More on topic, I don't believe we can necessarily say for sure that the miss rate is 27% just yet. Also, I don't believe we can necessarily break down the crit depression as you've suggested just yet. We could use some testing of the crit depression against level 81 and 82 mobs - which might be hard, since I don't even think it's possible to get a crit rate of 0.6% or even 1.6%. For a troll rogue, for example, the minimum crit rate with capped weapon skill and no gear, talents, or buffs is 2.00%.

However, there is nothing stopping us from running with the 4.8% crit depression against bosses, which is, after all, what we mostly use spreadsheets for. Although level 81 and 82 mobs do make appearances, and are certainly worthy of consideration, our specific knowledge of the crit mechanics against those types of mobs is not of vital importance since it won't significantly alter how we gear or play when fighting such mobs.

 

Demi9OD

AEP values for crit seem to be roughly equal for mutilate and combat, this depression may effect HaT builds the most, as they seem the most crit reliant (at least with the multiple CP bug). Does anyone else see additional effects this would have on rogue builds?

 

Custardcream

Well it reduces the number of focused attacks procs for a HfB build, so yes I can see these results having some impact (although probabally a minor one) on EP values.

 

sirann

Edited for retardation.

 

• Vulajin

I have no idea what you mean by this, but the tooltip is pretty clear that your off hand attacks, when successful (i.e. when they land), have a 20% chance to proc Combat Potency. In your case you landed roughly 8658 attacks and therefore expected about 20% * 8658 = 1732 Combat Potency procs, which is completely in line with your observed 1713 procs.

 

meneldor

Have done the same tests as most people here: took off all poisons, no special attacks and just let the autoattack run for a long time on the raidboss dummy (15 min). Recorded the results with Recount.

I use a slow fist in mainhand and the papercutter dagger in offhand, 186 hit, 22 expertise and a standard combat build.

As most people here I got some suspect results: my crit is reported in the UI as 30% but on the raidboss dummy I got a recorded crit-percentage of around ~21%. With my gear etc I would assume that I was missing
somewhere around 5% in crit?
(a few percent defecit should be ok since the raidboss has more defense than my 400 weapon skill)
First I thought that the boss had resillience or something

By chance I equipped another dagger in the mainhand and suddenly my reported crit-chance in the UI dropped to 25%!
Since I have 5/5 in close combat it should not matter if I have a fist or a dagger?
When running the bossdummy test again I still get the around the same ~21% in crit as before.

Seems to be a bug that when having fist equipped in the mainhand the crit-chance is reported as beeing 5% too much?


Sorry if this is a known issue?

 

• Vulajin

It is.

 

dabadguy

I can confirm that this isn't just effective against bosses. I ran as a 77 rogue vs the lvl 80 practice dummy for about 4000 swings, completely naked and observed a 2.2% crit rate compared to a paper doll crit rate of 6.9%. Don't know if you want me to rerun the tests as 78 vs 80 and 79 vs 80 to simulate 80 vs 81 and 82 mobs, but just leave a response here and I'll do it.

 

• Vulajin

Tests of the +1 and +2 level differences would be extremely helpful, provided that your weapon skill is capped at the maximum for your level in all cases. Was your weapon skill capped in this +3 test?

 

KraxisSingular

There are now some circumstantial evidence that Focused Aim for Hunters appear to force hits, locking out crits. This is still early, but if we suffer this, Rogues might suffer a similar issue?

Look from #33 and on.

 

• Vulajin

Not likely, since I experienced the 4.80% crit depression with a 0/0/0 talent spec.

 

Octaviann

I can do tests against +2 as soon as I hit 78 and then against +1 when I hit 79. What addon are you using to do the tracking? I should hit 78 in less than an hour or so from now and be able to put that data up later today. What crit percentage should I aim for, or should I just not worry about it and use my normal gear and then compare the expected crit rate to the actual?

Oh, and can you train weapon skill on the Dummies?

Edit: It looks like life is interfering and I won't be able to do testing, sorry.

 

• Aldriana

You can't level weapon skill on target dummies. I recommend the DM ogres for weapon skillups.

 

Octaviann

I was mainly wondering so that I know whether I need to go kill stuff for a few minutes before I go do testing.

 

Cleptor

While on the quest Securing the Bait in Sholazar Basin weapon skill points can be gained by attacking the primordial drake eggs.

 

FromTheAshes

This may be off topic but every spreadsheet ive looked at undervalues hit rating by a ton, instead favoring agility and raw ap. My understanding of the mechanics is if you cant hit, then you most def cant crit. And even as a mutilate build with focused attacks, it seems you would see as much as a gain in extra energy from hit rating and extra crit. Am i correct in assuming this?

 

• Aldriana

Attacks that don't indeed do less damage than ones that do. On the other hand, attacks that don't crit do less damage than those that do, and attacks with less Attack Power do less damage than those with more Attack Power. That is: all stats increase your DPS; the question is merely how much. Since figuring this out by inspection (as you're trying to do) is pretty much impossible, we use spreadsheets to model all the interactions. I assure you that all the factors you mention are included in the spreadsheets, thus we have reasonable confidence that the values in the spreadsheet are more accurate than whatever you intuition happens to say.

I'd also note that your assertion that an attack that does not hit cannot crit is misleading; in particular, increasing hit rating has no direct effect on the number of Focused Attacks procs you get - white attacks use a one-roll system, and, as such, your crit rate on white attacks is totally independant of you hit rate.

So, to summarize: you're wrong, in any of a number of different ways.

 

shegil

I did some dummy testing and got some interesting numbers, and I thought I'd share.

I used my standard gear but I swapped out my Murder with mongoose for Omen of Ruin, putting me at these unbuffed stats:

31.52% crit
267 hitrating (8.14%)
12 expertise (104 expertise rating or 3%)

and performed 12809 attacks on the boss dummy. Out of those attacks this was my result (used recount to gather the data):

3873 (30.2%) hit
3767 (29.4%) crit
3053 (23.8%) glancing
1707 (13.3%) miss
409 (3.2%) dodge

The expected amount of crits, assuming you have a critdepression of 4.8%, would be 31,52-4.8=26.72% crit. I had 29.4% crit, a 2.68% difference. I know that the sample size isn't that big, but enough for me to get naked and go to town with the dummy again this weekend, unless someone beats me to it *hint*.

 

• Aldriana

Well, the first thing I'd do is check the log to see if anyone applied a +crit buff - Master Poisoner or the like - to the target dummy at any point during your test. But if that's not the case, then yes, this would consist of evidence indicating that there's more going on than the straight crit reduction currently hypothesized.

 

shegil

Oh I'm such a douchebag, I'm specced into Master Poisoner for once due to lack of retri/ele. Nevermind my post, everything is in order.

 

FromTheAshes

Thanks Aldriana for setting me straight. I had a misunderstanding of the one-roll system with regards to attack tables.

 

DiracD

We should be able to just do a weighted average of all the tests to get a good total average of the miss rate. To get it to a decent confidence level where one could pretty much 100% assume the miss rate is 27% we would have to cut your 95% confidence range by a factor of 2 or so which would require about 80k total swings. The OP posted 11k swings on top of yours so we were missing about another 50k.

Then again those are just rough #'s off the top of my head. But the weighted miss average for 29463 swings is 27.101%

Edit: I'm working on some myself and I didnt weight some of the 1000 and 3000 swing averages in this thread because I am at work and didnt sift through their WWS.

Edit2: Went back through the post and picke dup the smaller averages and added ~12k of my own swings to get a average of 27.113 with 57462 total swings

My stats(tooltip stats):
using librarian's paper cutter and lightblade rivener
0Hit rating (3/5 precision)
10.89% crit (5/5 CQC with daggers)
0 expertise

results:
1435 attacks
25.2%miss
25% glancing
5.6%dodge
5.4%crit

7611 attacks
24.2%miss
24.3% glancing
6.2%dodge
6.5%crit

3500 attacks
24.3%miss
23.6%glancing
6.7%dodge
6.9%crit

 

Dontmindme

The best way to test the 27% hit cap is to load up enough +hit to reach this level.
Then...
a) With one less hit rating theoretically needed to cap display a miss.
b) With just barely enough to cap, show say 10000 attacks or so without a miss.

Barring the ability to do that, generally speaking the data would be better the closer one is to the cap. I applaud all the testing with 0 hit rating, but even a 10% deviation at 27% miss leaves 24.3% to 29.7%.

Whereas if one were even within 5% of the theoretical cap a 10% deviation runs 4.5% to 5.5% which more precisely defines where the hit cap really is.

In short, if you can, you would be better served loading as much hit rating onto your character as you can rather than taking it all off.