I still never stated this.
Really, mate. I was not joking before. You have three terrible habits based off this thread, You have a terrible habit of contradicting yourself, due to sentence structure or simply word choice, you exaggerate wildly, and you state thing that you feel, instead of what is actually there.
Do kindly stop.
I guess I would have to understand how the numbers are projected in order to evaluate the argument. For instance, is RNG included in projecting 99thile numbers? Any kind of naïve statistical procedure would conflate spec specific RNG variance with spec specific skill variance.
Is it the case that Warcraft Logs says a top 99th %ile Fury player will score X? Or is it saying that a projected 99th %ile Fury parse would score X? If it’s saying the latter then the 99th %ile is not really relevant as a balance point: all you’re doing is sorting the specs by RNG variance, rather than power. Fury naturally would do better in this kind of comparison.
If it somehow is saying the former, accurately, on the basis of 154 parses then… such miraculous statistical talent should be reallocated to more useful tasks.
Just put him on ignore. You will get a more fruitful discussion that way. I’d like to see some changes for sure, but I don’t think we’ll see anything until after the mythic race is over.
You linked the logs if 150 warrior to represent the entire class. That is highly misleading.
You did it to provide evidence when I said fury and arms are below average
True or False Derez
That is incorrect.
I do not believe so, but something more specific like that would likely only be known by Kihra. I try and keep myself relatively familiar with the site, but sometimes I’m taken aback by how complex and efficient it is.
The latter, if my understanding is correct. And it’s is not necessarily a bar to pass, as in, “You need to achieve at least X to be awarded with Y” but more, a general prediction, I.E., something around X will award Y.
I’m not sure I grasp your point, would you elaborate?
Did you reply to the wrong person…?
This is a comedy or errors.
No, I did not reply that to you stating that fury and arms are below average.
In fact, those words never came from you at all before now in this regard.
You stated that;
I specifically replied about the parts of what you were saying that were at best gross exaggerations, and at worst simply flat out inaccurate.
There is a significant difference between saying “Warrior dps is below average” and “One of the specs is dead last, and the other is near the bottom”
Do you understand where you went wrong yet.
What I said is accurate. Fury is dead last overall and arms is overall C tier.
Unless you choose to narrow your search down to 150 warriors instead of 10s of thousands.
Do you understand this is not how classes are balanced.
Still waiting for that post on your cutting edge main btw
Yes I am guilty of hyperbole. Does not change the fact that fury IS the worst overall dps spec and the other 2 are well below average
It’s easiest to explain in terms of one possible naïve statistical approach. A common way to project a 98th %ile number is to take the mean and standard deviation of a data set and then add two standard deviations. So say we have 100 parses with a mean DPS of 5000 and a standard deviation of 500, the (simple) assumption of a normal distribution outputs a 98th%ile DPS of 6000 (5000+500+500).
The problem with applying this (common) method here is that the standard deviation for a spec is a composite of RNG variance and skill (including gear obviously) variance. So if you have two specs with identical skill variance (the AVERAGE effect of being +2 SD above the mean in skill is the same) but different RNG variance, this method would show a higher standard deviation for the more RNG variant spec (manifested in the graph as a larger line between best and worst performers). This, in turn, would (dramatically) inflate the 99th%ile projections.
In other words all else equal (overall expected DPS, overall skill to outcome curve), specs with higher RNG variance will overperform at the top end of data constructed by normal statistical tools. A spec that averages 6000 DPS with perfect gear and play with no RNG variance will appear to be lower at the top %ile than a spec that averages 5500 DPS with perfect gear and play with a standard deviation RNG variance of 200.
He doesn’t just do it to warriors. He used to troll Warlock forums a few years back saying how Demo was fine even though it was a disaster.
I’m starting to think he’s a Dev’s alt account trying to justify why a specific class isn’t doing to well. Looking at Warcraft logs shows Warrior DPS is not doing that good, looking at M+ paints an even worse picture. They don’t have a spec doing above average to switch to like most other classes do. These are facts. He’s making excuses for these facts (or using 1 unrealistic data set, while nearly all others show otherwise) rather than refuting them.
I’m looking at the 90th, 85, 80, 75, 70 percentiles. Both warrior specs are below the mid point. This is not good. And lets not even get into M+.
Thanks I need to stop feeding him
It is not, no. You don’t even know how to read the chart you’re linking.
I’m familiar, and following you so far.
Ah, I understand your meaning now. Part of my confusion was, from this statement;
To clarify, fury is actually one of the more consistent dps specs at the moment, effectively tied with survival and assassination, and just slightly beaten by BM. I’m not sure how that translates to MT scenarios, but as fury’s rotational priority and stat priority typically stay mostly similar from increasing targets (or at least, sees far less dramatic shifts than most specs) I imagine the same can be said for it’s variance.
I do hope I didn’t misconstrue anything.
Can you quote that mate.
I specifically just showed that this statement isn’t exactly accurate.
What?
The top warrior dps is ranked 30 in the world for raw IO. Arms has timed 17 keys, fury 16s. Neither is a dominant spec, but to plaint it in this light is simply disingenuous at best.
How is that relevant.
https://www.warcraftlogs.com/zone/statistics/26#dataset=95
https://www.warcraftlogs.com/zone/statistics/26#dataset=99
https://www.warcraftlogs.com/zone/statistics/26#dataset=100
These are largely suboptimal players. Why would you intentionally choose to compare suboptimal data points.
You need to hold a coherent point, or when someone responds to what you say, not go wildly off course. Respond to what is said, not what you feel is said. You cannot read tone through text.
Lol and whats your point? The top 150 warriors have average parses in CE guilds.
Good i can strive to be slightly below average if I play perfectly and have access to endless resources.
Looks like warriors don’t need a buff bois
Not to mention this is all relying on the fact that you HAVE to choose venyth to be well below average in raid. The class has core mechanical issues that need to be addressed.
Yeah you may be right. I was thinking of Fury’s dependence on enrage RNG when I made that comment. Most specs don’t have RNG mechanics that powerful, but I’m hardly an expert on every spec.
I’m just speculating, though. I don’t actually know if Fury is more RNG dependent than, say, Unholy. My bigger point is that if these numbers are created by applying a standard deviation like procedure to a fairly small set of parse data then looking at only the +2.5 SD figures that result really doesn’t tell us very much. Both because that method will overrate RNG dependent specs and because the underlying statistical process cannot support meaningful predictions about the +2.5 SD case with an N of only 150.
It’s a bit like taking 100 mules with an average speed of 10 MPH and a standard deviation of 3 MPH and reasoning that the theoretical 99th%Ile mule will run at 17 MPH, and then using this as an argument that Jocky’s should race with Mules instead of horses because while horses run at a mean speed of 14 MPH the measured standard deviation of the horses was only 1, projecting that the 99th%ile Mule is slightly faster than the 99th%ile horse.
The math in the above is fine, the problem is just that the procedure is extremely dependent on the mean and standard deviation generated from the small sample size, as well as the assumption of a perfectly normal distribution. Perhaps the Mule SD is higher because a few of the mules were injured and came in at 5 MPH, radically inflating the SD. Perhaps our mules happened to be a bit faster than average. Maybe the mule speed curve isn’t really gaussian. The signal to noise ratio of this process is just not very good as applied to the +2.5 SD case.
It would be much safer to look at a broader if less exceptional sample of horses and mules. And if the mule mean is behind everywhere else it’s probably because mules really are slower than horses (including the fastest mule).
I have raider-io and warcraftlogs open in front of me. Warrior is below average in both, on the graphs or lists of both sites. There are classes with multiple specs that are on average better at both.
Are my sites bugged or something then?
Just for a quick point of reference;
https://www.simulationcraft.org/reports/T26_Raid.html
The dps comparison on this is largely unimportant (it’s actually a bit of a joke, which is why it’s not done every tier). The important bit is if you scroll down just a bit, and expand “Additional Raid Information”
Here you’ll get the two charts this is somewhat useful for, APM and DPS variance.
Only quoting this to save room, but in regards to the paragraph, I follow your reasoning, and it does make sense. For not having the data to go off of, you’re still at least staying on a consistent point and making reasonable steps. I really appreciate that.
Do you see how I gave specifics, and instead of responding to those, you give a generic response that is vague enough to refute me, without actually addressing anything I’ve said directly?
You mean, like the people they’re being compared to? Again, this is the second time you’ve done this. You’re implying that looking at the higher percentile is giving some extra edge to the warriors at that level, that is not available to the other specs. This is incorrect.
Again, at no point did I state this.
Actually, you’re the only one in the thread who has stated such, even if it was in a vain attempt to be mocking.
You really don’t though. What are you talking about?
The only weak covenant for either spec is necrolord.
Derez just disagrees to disagree basically. I know it was rhetorical saying carpal tunnel stems from playing fury and it does not scientifically guarantee literally getting it, but he had to state it was incorrect lmao. The joke is it’s a haste spam go go go spec that surely makes the keys pressing quickly for efficient gameplay.
I have the same apm on nearly any spec I play due to key spamming. This is rather common.
It’s not a funny joke. So I made fun of the joke. There’s a difference between being obtuse and taking the piss.
This is incorrect.
yeah yeah we get it buddy.
ps – do you live on these forums or what? go do stuff.
You disliking me personally does not require you to specifically aim to discredit me for no other reason than said disdain.
this is incorrect
Derez is absolutely dunking on everyone in this thread, and I hope it never stops.