A single sample size means nothing in this case because the proc is all up to chance and thus we could have seen a 70% proc chance on it and still then it would have been useless. Nothing to balance it out, nothing to weed out when you had extremely good/bad luck.
Everyone knows sometimes the RNG gods favor you, and sometimes they act as if you don't exist. That doesn't make a single test more accurate though.
- - - Updated - - -
Frost has very very little downtime even without taking that build that was mentioned, I imagine you will like it fine.
My point is, from that sample size (100) it is possible to prove something statistically (as much as statistics can prove anything). If I was to use obliterate 100 times and I got 99 rime procs, some might say: "sample size too small, you got lucky, rime is still 45% chance", but statistically the chance that rime is indeed 45% (in this scenario) is fucking zero (or close enough to).
Also an analysis of a single sample can indeed quantify chance. Also, especially in light of the fact that rime was recently bugged, I wouldn't be just assuming that it's definetly set at 45% now (again, I'm not saying it necessarily isn't).
We have blizzard saying specifically that it is now set at 45%. The reason it was higher before was it had more than one chance to proc per OB cast. It does not now, it has a single chance, so it's back to 45%.
Your argument only holds any kind of weight in a situation where we don't know these things, but we do, so I don't get why you're wasting your time fighting a battle that was never started and can't be won.
I don't think you know what my argument is. I have said multiple times that my argument is that just because a sample size is small doesn't mean the results are invalid, and a small sample size, after statistical analysis, has the potential to prove something. I've gone to pains to say my point isn't that rime isnt a 45% chance, I don't know if it is or isn't.
Blizzard said specifically that rime was a 45% chance to proc, but actually it was much higher. You appear to have concrete faith in blizzards ability to get everything 100% right, but obviously that not the case. That's why we, the community, can do these analysis, find out if a thing is actually working right, and do with that information what we will. Just because blizzard has said it is a 45% chance to proc doesn't mean it necessarily is ( again, note that I am not saying that rime isn't a 45% to proc, I am only saying that if there is data flying about saying its not, then we should be looking at that and not blindly saying that the data is wrong cos blizz says 45%).
I'll ignore the first part, because it means nothing.
They said back when Rime was added in initially, expansions ago, that it was 45%. It has been, 45%, since that time, and unless changed by a tier set bonus of otherwise that is how it has been. Assuming otherwise is silly, unless something consistantly shows up that proves otherwise. In general all Frost DK's were saying it procced a ton, someone dug into logs and found it was proccing three times per OB cast. That bug was then fixed back to 45%.
As of right now, there is no data showing that it's not back at 45%. One single persons experience, without logs, proof or anything other than words on a forum, mean nothing.
You're blowing things out of perportion to say "well it might not be 45% anymore" saying that there's data that says otherwise. Where? What data? How in depth was the testing? Was the testing repeated by that person or others and thus replicated to be true?
There is no data saying otherwise, but from testing you would see that, and obviously you don't care to see that.
go find a target dummy and cast Obliterate 1000 times and post a log
If you do this in a controlled environment and there is more than a 6-7% variance from the intended proc rate ill believe Rime is not proccing as intended. I think most would be more than willing to take that sample size Data (atleast reasonable people).
you would not even have to argue just post the log and it will speak volumes for you.
@Maxweii
1.) the first part doesn't mean nothing, it is the actual thing all my posts are about, which yes is counter to something that you said earlier. Small sample sizes can provide useful information, to the point that it can prove something.
2) There is no assuming going on here. The 18 out of 100 rime procs is a data point, you can describe it as poor evidence for a variety of legitimate reasons, but it is data.
3) FFS I'm not saying that this data proves that rime is not 45%. What I am saying is that because a sample size is small does not mean it is not valuable or that it can't prove a thing (which you implied).
This is an example of my point: if I were to flip a coin 100 times and get heads 100 times, statistical analysis (and common sense) would strongly suggest I'm flipping a double sided coin. I don't need to do 10,000 iterations, cos the chances that I get 100 heads in a row is so unlikely that favouring a different hypothesis, that being there is a head on each side of the coin, is the more likely explanation for the results I am getting.
For the last time: I AM NOT SAYING RIME IS LESS THAN 45%!!!!! I am only saying that dismissing data based only on its size is dumb, as it doesn't need to be big to prove a thing.
And I didn't say you said that, but now you're fixated on data that isn't there. Someone writing a post on the MMO-C forums means nothing without back up to it, there was no logs, nothing to prove what was said. I could have made a post in response and said "weeeeelllll I did the same test and got 89 out of 100 OB casts to proc Rime so your test means nothing!". And yet, even if I did without logs neither of what we said means anything.
Yet, when you critiqued the posters data you said it is not valid due to the sample size. You passed it on wholly and solely based on the sample size. This is what has prompted this conversation. Also I'm not fixated on this data, it is completely besides the point. My point is only that small samples sizes can prove something to not be true if the observed result is far enough away from the expected result.
You have absolutely no clue what you're talking about, so yeah guess there's a reason for some of the functions on this site. Least you can "sound" educated though.
? I am merely pointing out a really common misconception. No need to get butthurt.
Well, what we are talking about is actually a really basic stats principle. Significance can be reached if the observed result is a long way from the expected result even at low sample sizes depending on how far apart they are and the sample size. This isn't like a thing that is arguable, this is fact. But hey, continue to be ignorant if you want, I can't stop you.
Logs from Xavius tests:
https://www.warcraftlogs.com/reports...done&source=20
SF seems to have an issue hitting slow immune mobs,
Positioning:
https://www.warcraftlogs.com/reports...j4#view=replay
Immune log:
https://www.warcraftlogs.com/reports...91&end=3627255
This is the only attempt I care to dig up, as its the the one we decided to kill it, But it happened on almost every attempt.
(Also didn't seem to hit Xavius multiple times as you would expect on a larger model, Possibly fixed)
Frost seems to still be far and away the best ST spec, The Feral was close but she rarely did soak mechanics even with the Dream buff, so she essentially tunneled the boss the entire time.
Nerf Hunter Execute. =)
Last edited by Nemesquish; 2016-07-11 at 03:19 PM.
Keep in mind ambidextry bug still affect the Main-Hand Component of FS and Oblit.
\
Yea, Even with that in mind, Only the Feral would have beat/been close on ST damage, And I didn't tunnel.
The RL and I actually had a talk last night after raid, since DK mobility is so bad, and some of those pools have to be soaked ~40 yards away, I'm just going to tunnel next time we do this, Then see the difference.
(I'll still do close pools, cause I'm by no means a scumbag, but ill skip the 10 second downtime pools lol)
Sindragosa's Fury doesn't hit large models more than once it only hits targets more than once if there's multiple present in the hit area. No matter how large it is if it's a single target it will only be hit once.
Is remorseless winter supposed to be multitarget only? It does like literally 0 damage.