[Bug] Berserker Stance


#4

White hits are one-roll. We can verify this by noting that with enough samples, all the outcomes match their expected probabilities.

Yellow hits are two-roll (for player attacks on mobs). We can verify this by noting critical block events.

The tests we did concerning berserker stance missing crit vs +3 are white hit tests. Again, we are getting expected results for same level mobs, but for +3 we are missing 6% crit when we should only be missing 3% crit.


#5

I should add that two-roll is a shorthand meaning certain outcomes can overlap (e.g. crit and block) and it also means outcome chances don’t all match up with tooltip values (such as crit). It doesn’t mean you need to actually do two random rolls, it can be done with a single roll and special interpretation but the implementation details aren’t important. Again none of that is relevant to the berserker stance bug since we’re talking white hits.


#6

Please fix this, th anks.


#7

good work gamers. Hopefully blizzard looks into this!


#8

Please fix this, th anks.


(Aggrend) #9

Hi All,

We looked into this.

Modifiers to Critical Strike chance are reduced against enemies 3 levels higher than the player-character. We tested a variety of modifiers to Critical Strike, such as: Berserker Stance, the Cruelty talent, and critical strike chance on gear. They all behaved the same as they did in 1.12, so we’ve concluded that there’s no bug here.

Thanks for the report!


#10

To be clear, you are stating that berserker stance should give 0 crit when fighting +3mobs? This is on top of the 3% crit supression already known


(Aggrend) #11

I’m not going to get too deep into the specific math on this one, but for the sake of clarity, I’ll say that this is not a per-aura modifier and it is separate than the 3% suppression that is baseline against +3 level creatures. There is a single modifier placed on total Critical Strike gained from auras that increase Crit chance such as those I listed in my last post.


Crit Auras and Raid bosses
#12

Thank you for the response, that clears up a bunch! It took a long time to build the hit tables used in the OP so its appreciated that you responded and that it wasnt all for naught.

Is there any chance you could say what the actual number for the “single modifier placed on total Critical Strike gained from auras” is? Is it always -3% or will it scale with how much additional crit you gain from auras?


(Aggrend) #13

I can imagine. It was really important to us that we get some sort of response to you before the end of beta.

Nope :slight_smile:


#14

Would you be able to say if this formula existed throughout vanilla or was added later on?


#15

Looks like people will have to do a bit of science when the game comes out to figure out exactly how the math works, then.

It’d be weird if we found out the crit talents were suppressed and thus weren’t worth nearly as much as we thought, but oh well.


#17

Wow there’s a lot to unpack in this thread…
Seems like warrior gearing BiS could change since the crit% soft cap will be some unknown number higher than 33% now. To ensure flurry has a 100% to near 100% uptime.
Furthermore it looks like its an additional 1% per mob level reduction of crit from aura buffs. Warrior soft crit cap 36%?


#18

For anyone interested, we’ve done some extensive research about this since Blizzard’s response which can be found here:

https://github.com/magey/classic-warrior/wiki/Crit-aura-suppression

I’ll quote the TLDR:

Weighted average by number of attacks for all suppression results (without including the 1% aura crit tests which seem to be suppressed entirely) gives ~%4.84 average suppression with a 95% confidence interval of ±0.17%, meaning the suppression to aura crit is likely ~1.8%.


#19

Isn’t it more likely that the crit suppression is a percentage of the additional crit, rather than a flat number? So for higher value of crit, the suppression is greater.


#20

We collected over 60k hits from 10 different testers to arrive at that result. Isn’t it more likely that the theory based on 60k data points is more correct than the theory based on 0 data points?


#21

Awesome work, has there been any testing with casters? Is the operating assumption that the reduction is the same flat % for them as well?


#22

When your statistical methodology is wrong, I would say a theory based on logic is more likely to be correct. I took the liberty of doing some proper analysis.

We know that there is a suppression on crit auras, so a proper comparison would look at the additional suppression above 3% vs. the amount of aura crit, and then fit a linear regression to that to arrive at three possible conclusions.

  • There is an intercept and zero (or close enough) coefficient on the aura crit, which means aura crit is suppressed by a flat amount

  • There is an intercept and a coefficient on the aura crit, which means aura crit is suppressed a flat amount plus some percentage of aura crit

  • There is no intercept and a coefficient on the aura crit, which means aura crit is suppressed by a percentage of aura crit (What I believe is most likely).

I fit two regression lines, one with an intercept, and one without, the results are as follows:

  • With intercept: Suppression = 1.58% + 0.0228 * aura_crit (95% CI of ±0.003% on the intercept, the slope is not statistically significant)

  • Without intercept: Suppression = 0.1770 * aura_crit (95% CI of ±0.0062 on the slope)

I can provide the Python notebook for review if you’d like, but what this tells us is that we don’t have nearly enough data to determine which is accurate, as you might think it’s 60k data points, but in fact is it only 16.


#23

Unfortunately there was very limited time to do further tests, and we were only able to get a couple of caster logs, one from shaman Ayle and one from a druid. The shaman log appears to show suppression but the druid one doesn’t. Don’t read anything into that, could just be sample size so we just don’t know yet.


#24

Ok I’ll try one more time cause this is very basic. Let’s go back to probability 101 and roll a dice 1 million times, then roll a dice again 1 million more times. You are arguing the sample size of that experiment is 2. But of course in reality the sample size is 2 million. I grant there are some subtleties in aggregating WoW combat data, but saying our sample size is “16” is pretty obviously incorrect. I could take one of those logs and split it in half, and by your logic I’d have 17 samples now. I think you can see the flawed logic there.

As for the data itself, check the suppression column in our data and you won’t see an increase, which there would be if it were percentage based. E.g. our test with 17% crit aura showed a crit aura suppression of 1.91%. Is there a chance the suppression is actually 10% for 17% crit aura? Sure but the chances are remote in the extreme. Scroll down to see the graph and as you can see, it’s pretty flat.

If you need more help in seeing where your code went wrong, post it on our github and we’ll have a look.

Better yet, you can test this yourself on the next stress test. Get 5/5 cruelty and go to town on a +3. Respec and do another test with only 1/5. Post your combat log to our github and we’ll happily review it.