I went back to the hit rating thread, where people had posted screenshots of their results. The screenshots show the character sheet info as well as the Recount window, so although the tests were for hit, you can see the crit results as well, which I have summarized below.
All the tests were done on Heroic Practice Dummies, and all had at least 2000 shots. All results are based on Autoshot or pet Melee unless otherwise noted.
The pets' recorded crit varied a lot more than the hunters' probably because the pets' hit ratings were more various. Some of the pets "over-performed" which may be due to the larger sample or another factor.
EDIT: It was pointed out below that the pets' spec was the most likely reason for the larger variance in pet crit rates -- those clustering around 20% and those clustering around 10%. I concur. My own pet tested for a 20.3% crit rate, and he is a wolf (ferocity).
The hunters' results are relatively consistent, however, with mean 5.3% depression compared to the Paper Doll stat, and median 5.7% depression.
Last edited by Rosamonde : 12/09/08 at 4:10 PM.
Reason: Remove Paper Doll Crit and Difference columns for pet data
The other factor in crit variance is probably due to there being more points in cobra strikes, thus at 3/3 cobra strikes, the pet will crit more often on specials then the hunter will over all.
It would be proper to test pet crit without any points in cobra strikes. or to test pet crit without the hunter firing, so as to not trigger cobra strikes while firing steady shot . In my test I was hit capped without FA, so my pet had no misses, and his crit rate was also depressed. You will probably find that in the test in which hunter pets had the highest returns, the hunter was also firing steady shot, or another special to trigger cobra strikes.
Also 2000 is a good test sample, but it is still subject to variance. I was going to stop at 2000 shots on the first test, but I observed excessive crit variances per 100 - 200 shots and stopped testing after I only saw a only a .1 percent variance over an extended period of time.
I stopped at 2000 shots on the level 80 test dummy because I was close enough at that point to be able account for variances, and there is no way to make up a 4.75 (4.8 as base) over the next 2000 shots, unless you had a 9 percent OVER tool tip crit return.
Could you link these test so others can to extrapolate some data from them?
I probably should look them up myself, but I am watching Monday Night Football. Go Panthers!
Is it a coincidence I am a football stat buff as well....
Rosamonde, pets don't inherit the Hunter's crit so you can't draw any relation between the two. And Claw should really be disregarded unless the Hunter has specifically mentioned they don't have any points in Cobra Strikes.
A pet with Spider's Bite and a BM Hunter should have 24% base crit. Let's work from that, again unless specifically mentioned to be otherwise.
Well, the question was brought up whether crit was behaving differently than expected, gauging by the figure in the character sheet. I thought the data we had already accumulated in the other thread might be useful -- I make no claims about it!
If we disregard the pet data, the hunter data clusters around what Thayer found in his original test -- an average of 5.3% less than what is stated on the character sheet.
These tests were all done after last Tuesday's (Dec 2) maintenance as well. If I get bored at work tomorrow I will dig around and see if I can find any posted results that were done before then.
Well the pets tend to sneak in in two groups. One around 19%, and one around 9% crit. That fits a Ferocity and Cunning/Tenacity setup pretty well. But the divergence is bigger for pets, I agree.
MM dwarf with a gun vs. lvl83 boss target dummy (35.22% crit tooltip).
Screenshot
70.9% hit
29.1% crit
= 6.12% crit depression from tooltip, really strange.
(Note that I got 9.18% WQ instead of 10%)
I was noticing about 6.2 crit depression when I had Lethal shots and Focused Aimed as talents, which lead to my first assumption that somehow lethal shot was not being calculated (5 percent - 1.2 percent old assumed crit depression). From the paper doll it indeed looks like you had Focused Aim as a talent (162 hit). Try the test again with hit only provided by gear and see what you get.
It may be possible that the hit provided by Focused Aim is somehow unable to crit.
I have been meaning to test these as factors, but real life demands attention as well.
Adding in Thayer and Nukk's results gives a sample size of 11.
Mean crit depression in the sample is 5.47, and median is 5.75.
With a standard deviation of 1.06, that gives a range of 4.41-6.53 less crit in practice than the character sheet shows.
Since a lot of these tests were intentionally done with less than 8% hit (because we were testing hit), the sample is not very clean.
I set up an AFK target dummy test before I left for work this morning using 8% hit from gear, 0/3 FA, and 5/5 Lethal Shots. Tonight I will retest with 0/5 Lethal Shots.
It will be a lot more tedious to test Steady Shot since you have to stay at the keyboard the whole time...
Rosamonde, the variance seems to be due to Focused Aim. I will post some result shortly.
I try not to AFK test, since paladins have a pesky habit of running up and putting the 3 percent crit debuff on the dummy. So yes, I sit and stare at my avatar for hours at a time, alt tabbing frequently to type a bit and back.
Here is my first test on a level 83 dummy, with 3 points in Focused Aim
6.82 percent less crit then the tooltip, or about 2 percent more crit lost now that I have Focused Aim. This seems like a very heavy loss for getting 2.78 hit from FA (my gear hit was 5.22).
I am not going to make determinations until I finish testing on a level 80 dummy to see if there is any crit loss.
6.82 percent less crit then the tooltip, or about 2 percent more crit lost now that I have Focused Aim. This seems like a very heavy loss for getting 2.78 hit from FA (my gear hit was 5.22).
I am ot going to make determinations until i finish testing on a level 80 dummy to see if there is any crit loss.
Focused Aim just keeps getting better and better huh? I hope you are seeing a wrong sample of sorts as this is just very poor design. But thinking about it, it would make sense. Focused Aim forces a hit, so if it must hit it can't crit. Maybe the mechanic behind hit talents has changed from a missremover (lowering miss essentially) to a hit increaser, which would fit the result just posted.
Focused Aim just keeps getting better and better huh? I hope you are seeing a wrong sample of sorts as this is just very poor design. But thinking about it, it would make sense. Focused Aim forces a hit, so if it must hit it can't crit. Maybe the mechanic behind hit talents has changed from a missremover (lowering miss essentially) to a hit increaser, which would fit the result just posted.
Yup. I am testing on the level 80 dummy, and I am already seeing about 1.7 percent less crit then tooltip.
I lost 2.18 percent crit over 2000 shots. Since I have 5 percent Hit from gear, I'm running another test on the level 80 dummy in the same gear, without Focused Aim.
I am going to have to run large test sample sizes for these as well, as I watched RNG roll in and take it from a -0.3 percent tool tip difference to a -1.2 percent tool tip difference in just the last 300 shots. I will note my pet crit 24 percent, so I think he was stealing all my crit rolls.
So far it seems that without Focused Aim I am still critting 1- 1.7 percent more then with it. I'll have to run a larger sample to stabilze this, but I will have to do it later, as I have to raid sometime.
So far we have a lot of reports of our effective Crit ratings being lower than our paper doll ratings.
What no-one seems to have tested so far is if it makes any difference if we have the talent or not.
In order to test that, I'll conduct a test tomorrow myself, and would love others to try the same. First I will do a session of 2000 autoshots without the talent (actually without any other talents). Than I'll do another session with 2000 shots with the talent. Comparing these two will prove if it is the talent that makes us lack the 5% crit or not.
A similar test could also be ran of Steady Shots (and other shots).
So far we have a lot of reports of our effective Crit ratings being lower than our paper doll ratings.
What no-one seems to have tested so far is if it makes any difference if we have the talent or not.
I am running such a test even as I write. I will post the screenshots as soon as I finish the second phase of the test, but thus far it does not appear that Lethal Shots is the culprit.
My results confirm what Thayer posted.
Heroic Test Dummy -- 3000 Autoshots
0/3 FA 5/5 Lethal Shots
274 (8.36%) hit rating from gear
19.69% Paper Doll crit
13.9% Actual crit
5.79% Crit depression Screenshot
Heroic Test Dummy -- 2000 Autoshots
0/3 FA 0/5 Lethal Shots
274 (8.36%) hit rating from gear
14.69% Paper Doll crit
9.3% Actual crit
5.39% Crit depression Screenshot
Last edited by Rosamonde : 12/09/08 at 10:01 PM.
Reason: Add screenshots
So far we have a lot of reports of our effective Crit ratings being lower than our paper doll ratings.
What no-one seems to have tested so far is if it makes any difference if we have the talent or not.
In order to test that, I'll conduct a test tomorrow myself, and would love others to try the same. First I will do a session of 2000 autoshots without the talent (actually without any other talents). Than I'll do another session with 2000 shots with the talent. Comparing these two will prove if it is the talent that makes us lack the 5% crit or not.
A similar test could also be ran of Steady Shots (and other shots).
All my recent tests were without the talent. If it was lethal shots, it would have to make up 10 percent crit all together, since taking lethal shots reflects on the spreadsheet and on performance.
Originally, my first set of tests were with the talent, and it is where I noticed the crit reduction first.
There is always a 5 percent or so crit reduction with or without the talent.
We already covered that it was not lethal shots on the first page. Right now we need to test to see if taking Focused Aim somehow reduces our crit performance as well.
Originally Posted by Rosamonde
I am running such a test even as I write. I will post the screenshots as soon as I finish the second phase of the test, but thus far it does not appear that Lethal Shots is the culprit.
My results confirm what Thayer posted.
Nice. I think if we test Focused aim we need to do around 4000 - 5000 shots, since at 2000, and even 3000 shots, I am still seeing RNG effects of up to 1 percent, and since the effect of Focused aim is around 2 percent as it seems, we need to reduce the tolerance for error when testing to see if this is also a factor, and to see exactly how much of a factor it is.
All my recent tests were without the talent. If it was lethal shots, it would have to make up 10 percent crit all together, since taking lethal shots reflects on the spreadsheet and on performance.
Originally, my first set of tests were with the talent, and it is where I noticed the crit reduction first.
There is always a 5 percent or so crit reduction with or without the talent.
We already covered that it was not lethal shots on the first page. Right now we need to test to see if taking Focused Aim somehow reduces our crit performance as well.
What I was referring to was exactly what Rosamondes testing confirmed. There was no test done aimed exclusively at confirming if Lethal Shots had any hand in this.
Doing the tests without Lethal Shots might have come up with less of a Crit than our paper doll showed up, but there was no direct comparison yet with doing tests with Lethal shots. Rosamondes test showed exactly what my question was aimed at. Whether or not Lethal Shots had any hand in this, to the exclusion of Lethal shots.
Just going to bounce an idea off here, but did you look at the actual damage dealt by your crits? Is it possible that they could have added 400 resilience to bosses? This would lead to an almost exactly 5% reduction in crit rate, but also reduce crit damage by about 10%. If this is the case, it should be pretty easy to recognize -- your average crit being only 1.9x after a healthy sample size would be pretty noticeable.
Rosamonde, the variance seems to be due to Focused Aim. I will post some result shortly.
I try not to AFK test, since paladins have a pesky habit of running up and putting the 3 percent crit debuff on the dummy. So yes, I sit and stare at my avatar for hours at a time, alt tabbing frequently to type a bit and back.
Go to Darnassus there is never and i mean never anyone there.
Just going to bounce an idea off here, but did you look at the actual damage dealt by your crits? Is it possible that they could have added 400 resilience to bosses? This would lead to an almost exactly 5% reduction in crit rate, but also reduce crit damage by about 10%. If this is the case, it should be pretty easy to recognize -- your average crit being only 1.9x after a healthy sample size would be pretty noticeable.
Looking at the images sunflame posted, the crit dmg rate on the horic boss dummy was 87.15% of the lvl 80 dummy. However, the non-crit rate was also down between the two dummies, meaning that you can't assume that resiliance is the cause, e.g. there could be armor differences between the two dummies.
Last edited by Iru : 12/23/08 at 1:57 PM.
Reason: forgot a contraction
Paper Doll crit 18.98%
Auto crit = 19.5%
Steady crit = 19.5%
Avg Crit = 19.5%
0.53% Appreciation of crit
2)
5/5 Leathal Shot
0/3 FA
Paper Doll crit = 23.98%
Auto crit = 21.2%
Steady crit = 24.6%
Avg crit = 22.9%
1.08% Depreciation of crit
I do believe testing on the heroic dummy will always give lower crit rates because it is a few levels higher, so those test should be viewed skeptically.
12/08/08, 8:01 PM
Similar Threads
