Hardiness percentage is not correct

So, to test it out without much speculation, I asked an orc shaman to stand still in blasted lands and afk for 1.5 hours. During this period, I casted 100 hammer of justices on the orc.

59 resisted (59%)
41 landed (41%)

You need to look into this Blizzard. Your numbers are not correct. I decided to test this out right after another orc resisted 6 HoJs in a row during a black lotus spawn fight in EPL.
There is no way the resist chance is 25%(+5%). Any alliance pvper with some sort of stun will tell you it is easily 50%(+5%).

Look into it and fix your numbers to make the game a bit more bearable. Thanks.

33 Likes

I totally stopped trying to stun orcs with openers from stealth. I was keeping count of the resists, and from 20 openers I landed only 4. Which is not that far away from your own findings.

10 Likes

Question, was said orc perhaps a warrior with iron will also?

As an orc myself I have never resisted more then 1 stun in a row that I can remember ontop of my head.

2 Likes

Make a video, no1 trusts random words.
Yet i agree orc resist is obviously bugged.

6 Likes

First, a very nice topic. And also, cheers for the test that you performed.

However, based on my professional experience (I am a software quality assurance guy) I can tell you that 100 tries is not presentable statistics. This number is too low. In order to have a correct observation, you have to do at least 10 000, while 100 000 is the first number that might yield relevant results. Of course, only Blizzard has these values, as it’s not practical to simulate this in-game without the use of a simulator. As are all claims that the actual value of the stun resist is not correct.

15 Likes

It is not a great stat, but it is the best we got. It reinforces a suspicion that MANY people have been voicing.
Ofc we need more numbers and more people doing it, but it is enough to raise concern.

As a personnal experience, i’ll only say that I specced my hunter out of bm-stun because it was to unreliable. Sadly I have numvers to show.

1 Like

If those numbers are true that is pretty sick. It’s a big N, so you might even think its no coincidence

While obviously a tiny sample size, it DOES echo my anecdotal experience of literal hundreds of “charge resist, intercept resist concussion blow resist”. The staggering amount of triple resists I’ve seen makes 30% stun resistance seem very unlikely.

I had an especially notable WSG where I failed to stun orcs entirely. Did not land a single charge or intercept stun on any orc in the BG.

4 Likes

Well of course not, once you start ghosting and never reviving after 1 or 2 deaths you obviously ain’t stunning much. :joy:

3 Likes

he said orc shaman what

1 Like

You should stun another race 100 times to get some data to compare it to.

Assuming a base stun resistance of 0%:
0.75 * 0.98 * 0.99 = 0.73 should be ur sucess rate. (1-stun resist) * (spell miss chance in pvp) * (flat 1% miss for spells)

btw spec repentance if u are gonna pvp, much better than holy shock

1 Like

Uhm… why not? Assuming the data is correct and not manipulated, this is absolutely VERY significant. It’s not impossible that 30% is still the actual resist rate but it’s extremely unlikely. If we were to reject a resist rate of 30% with 59 successes we’d be wrong in <1% of all cases.

5 Likes

Ah sorry I missed that my bad, stil the sample size is too small even with the % chance its still RNG ontop of it, would need a minimum of 5000 tries at least to get reliable statistics.

1 Like

The sample size is absolutely fine… As I said, with that data you can be almost certain that p=0.3 is incorrect.

5 Likes

I flip a coin 100 times.

Now statistically it should be 50/50 of heads vs tails … but I could get all heads, all tails or 75 heads / 25 tails. It is all probably and only when getting to larger statistically viable samples does the actual and most likely probabilities become apparent.

Or a little simpler - i toss a coin 10 times, it all lands on heads (stun) … does that mean it is 100% chance to stun? Of course not.

7 Likes

Ok, please tell me… what is the likelihood of getting >= 59 successes in 100 trials with a success chance of 30%?

5 Likes

I was about to say this! ^

About the same chance as when you flip a coin and it lands on its face 7.5/10 times.

3 Likes

The chance is about 0.00000018%.
The chance to flip a (fair) coin and hit 7 or more faces in 10 tries is about 17%.

8 Likes

I think it’s better to answer to statistical questions with statistical answers. While I agree the number of trials isn’t huge, it’s possible to compute the p-value and likelihood and all that fancy precise stuff, instead of throwing random numbers like “you need bazillion trillions observations to be sure”.

I’d do that myself but I’m too dumb, but even I know these things are possible to discuss rigorously.

5 Likes

Thank you! 100 isn’t the biggest sample size, but absolutely enough to reject 30% stun resistance with very high certainty in this case.

Assuming the data has been collected properly of course…

8 Likes