It was stealth nerfed for Arcane sometime recently and we only discovered it a day or so ago; so re-sim your minor essences with simc nightly build if you’re playing Arcane and see where Lucid Dreams minor now ranks for you, this only affected Arcane, Fire and Frost haven’t been changed. Crucible just got a whole lot more desirable for Arcane.
https://cdn.discordapp.com/attachments/325694580584349707/651505833741516800/unknown.png - The proc rate is now 7.5% down from 15% for Arcane.
Dangit! That sucks!!!
Makes sense. Arcane mages were over performing across the board …
… and please read that in the sarcastic tone it was typed in.
…why? I feel like Blizzard hired a professor who sits around theorizing about these things without doing any field work.
Crucible/Strife minors are better anyway, it’s not really a nerf to Arcane as we had other equally good options.
Oh no. 
Can’t we just have a nice interaction with an Essence for once?
They’re better by comparison now, but it looks like the top arcane DPS are still using Lucid/Strife. Quick glance through recent logs shows the top Crucible/Strife combo (Nov 17th) about 5k lower than the top Lucid/Strife combo (Nov 27th) on both Sivara and Ashvane. If the nerf happened after Nov 27th, I’m not sure it would be significant enough to make Crucible/Strife an automatic go-to.
Isn’t Strife the pvp essence? Another reason not go back to arcane: no interest in pvp.
Haha, if you have no interest in PvP then arcane is the spec for you. Strife is the PvP essence, but the minor is just a simple stacking versatility buff. Good for all-around use, and it offers comparable DPS to crucible because of arcane’s lower proc rate/uptime for the DoT. Mentioned this in another thread, but if the CD reduction on the VoP minor scales to at least 30%, I wouldn’t be surprised to see that on a lot of 8.3 arcane mages (we’ll have three minors).
It happened sometime in the last few weeks but we only found out about it a few days ago so not even sim data knew about it so it was being evaluated incorrectly; sims have been updated now though.
This is an incredibly stupid way to compare.
Is there a rule that posters on this forum have to be exceptionally rude? If you wouldn’t talk to someone like that in person, don’t do it here.
No, it’s not a bad way to compare, it’s the only way to compare that makes sense. We’re talking about build potential and the best way to measure potential is to look at the top numbers, not the average. It wouldn’t make sense to say that Crucible/Strife is a better combo if people are still regularly breaking its premiere DPS benchmarks with Lucid/Strife or Lucid/Crucible.
This is literally how I talk. Don’t make assumptions, something about you and me.
Yes it is, rofl.
lol?
It wouldn’t make any sense to try and make assumptions based off r1 parses. Especially when; the change is relatively recent, it went unnoticed for an extended period of time, there’s likely a decent number of people who still don’t know about it, God knows what the asians are aware of, etc.
Let’s assume they’ve been fully aware of the change since it occurred, and let’s also assume it occurred 4 weeks ago. Let’s make a random number and say of the 478 arcane parses on ashvane, 400 are asians. Now let’s totally realistically pretend that all of them were in on every kill and never once sat for farm to let someone else go in, or felt like playing a different spec/alt.
That’s a maximum of 1600 kills. Now, what do you think the chances are the factors of the raid are the same across all four kills, completely disregarding individual RNG of arcane itself, (which is idiotic itself considering arcane is in the upper percentile for performance variance)
That’s a ridiculously small sample size, and that’s making all the best possible assumptions to make it not stupidly small.
Do I need to go further into why this was an incredibly silly observation to make.
Then you have much more serious problems than whatever is happening in this thread.
It’s not a small sample size. It’s looking at the entire set of results from a sample size of ~1500 records over two bosses, which is easily and significantly over a reasonable threshold to draw conclusions from. Not sure why you’re randomly dismissing Chinese records with no sound explanation. I specifically qualified the dates of the entries in my post, so I don’t know what your problem is other than trying to pick a fight.
A sample size using factors that I deliberately (and openly) fabricated in order to try and make it not-stupidly small.
And really. A “good” sample size is something in the multiple thousands, at least, when it comes to logs. There’s a reasons sims run a minimum of 10k+ iterations, and usually more. Sometimes as high as over 100k, depending on variation in the sim itself.
This is the same as looking at logs the first few days mythic is out and trying to draw a conclusion off of that.
…what.
I’m not sure what you’re trying to say. 1606 valid records is 1606 valid records. That’s our sample size.
I don’t know what statistics program taught you that, but it’s terribly terribly terribly incorrect. A minimum acceptable sample size will be dependent on what exactly you’re trying to infer and for what population – it could be as low as 30, in the most basic cases. A sample size of 1606 relative to the population of moderately well-geared arcane mages in EP will handily meet a 95% CI with room to spare.
The reason is because they can, and because it minimizes the MoE. While that would be preferable, we don’t have 10k+ records of real arcane mage performance on hand. We still have plenty of data to draw reasonable expectations of how we can expect arcane mages to perform.
Yes but we’re not dealing with normal statistics, normal statistics aren’t subject to incredibly high amounts of the variant rng you find in a video game
…ok. Pretty much anyone who’s been doing this for a while, least of all me, would disagree, but I can’t fix everyone.
Of course they are. There’s no “abnormal” statistics we use for sets of data that are arbitrarily exempt from basic math. It’s not like real life variables are more predictable than video games. The fact that video games use RNG doesn’t mean anything, what matters is how significant the variation is. Higher variance data sets require larger sample sizes to meet CI thresholds, but not a single reputable statistician is going to say that a sample size of 1606 is too small to get a relatively accurate reading of damage output expectations.
If 100,000 sims say that Trait A is better, but a sample size of 1,000 logs says that Trait B gives better numbers, you should be trusting the logs, not the sims. It means that there are (most likely) variables that the sims aren’t accounting for.
They are, in a controlled environment. Which is usually what you’d take statistics from. And it depends entirely on the scenario. If you really want to go into a stupidly broad terms, then it depends. But usually you’re not gathering statistics for things that vary by such degrees. There’s just too many layers of rng and variance in a video game, and you should be aware of that if you’re really going to talk about statistics.
I mean, it does.
Certainly, assuming there’s some commonality or control. The only consistent factor is arcane.
Alright, so go into the sims and find that for us, aye lad?
just…Christ, no.
In response to a simple explanation of basic, high school-level theory of inferential statistics, you took 200 words to say “nuh uh.” Good talk.