The funny thing is that they also do exactly what you did to look at the “skill trajectory” of a deck.
These aren’t just useless numbers that you can only look at and say “neat, those sure are numbers!”
You know this isn’t true, so you’re being disingenuous here at best.
If you can tell me exactly how much of that made up crap is skill and how much is other factors, do it. If you can’t, then you are guessing.
And I have said it is a non-zero factor repeatedly through this discussion. But you know, and I know you know, that this analysis doesn’t yield any sort of accurate or reliable measure of skill. Letting them believe otherwise is lying to them.
Show me where they did what he did above. Link it or you are just talking out your backside. They did not do “exactly” anything.
Side note: The text you quoted in your post from the person I have on ignore explains why the numbers in the first post don’t match, the numbers he calculated based on games played. It isn’t so someone can’t “steal their numbers” it’s literally an artifact of how they calculate the weighted average. But more importantly, the reason you can’t just go from one meta to the next like they want to is precisely because of how weighted averages are calculated - the numbers exist only because all of the other conditions exist, a concept that is lost on the op.
… I did. Repeatedly.
Their math is correct. I don’t argue that there is some calculation error.
The whole point is they took descriptive statistics and tried to turn them into inferential statistics, but lack sufficiently good underlying data to do so.
They are measuring something, but what they are measuring is not useful, accurate, reliable, or truly helpful.
Scrotie said it was all skill (since corrected). You’ve said it’s none. Both are low probability guesses.
I’m not guessing, I’m not putting any number on it. As far as what the skill value is, I’m saying the balance of probability is that “skill” is not zero, simply because there are many more combinations where the skill value does not equal zero.
My only critique of the original post is that I don’t think the sample is sufficient, which is basically what you’re saying just without all the insults and with a couple of sentences.
Also for what it’s worth, it would be impossible to prove a zero skill hypothesis with 100% precision.
1 Like
The point made about the quality of the data is true if you want to do a serious study or something BUT…
For simpletons who not need the incredible exact number but are trying to identity stuff like meta trends instead it’s “good enough”.
To be fair we could even do an argument about not even blizzard itself having “good data” so it’s better to just go with what we have.
With that said:
It’s just a card game.
Why there is always that incredible necessity of all types of acusations coming from neon?
I don’t know what this game done to him but for sure it did hit harder than he will ever admit.
I have not said it is zero skill, and you are intentionally misrepresenting my position.
I said you can’t reliably tell what part is skil and what is other factors, making the whole thing useless.
You know that’s correct.
You are not putting a number on it because you can’t. That’s the whole point.
This number isn’t “skill” in any realistic way. Skill might be a part of it, but so is deck changes, meta changes, rounding error, correlated error terms, and every thing else - all the signal and all the noise with no dial to tune it in.
No, you’re out here acting like you disagree with me to help their feelings instead of telling them factually that calculations of this kind from samples this poor are unreliable and generally frowned upon because they don’t accurately measure what they think they do.
I challenge you to re-read where I say the higher bracket has more skilled players, but we aren’t measuring that with these calculations. I have used the words axiomatic to describe that more skilled players are at higher brackets.
THe only think I have rejected is the op’s “measure” of skill as being actually a measure of pure skill enough to be useful.
And you don’t see the irony of this being an accusation?
Let me translate your post: You’re right, Neon, but these people are:
so just let them be.
The game is fine. The op has literally told you his whole purpose in making this post was to try to prove me wrong, yet it’s now my fault his garbage stats are garbage.
You guys should all just tell him plainly that he should go back to driving his uber and leave the stats to other people.
1 Like
You can eliminate most of the other factors, what’s left over implies skill.
It’s not hard to notice that a deck has improved matchups as it climbs from diamond to legend to mid legend to high legend. Even with the sample sizes becoming less reliable, there’s only reason to question the validity of them if they were significantly breaking the trend.
You can also track deck popularity, and with simple algebra determine how much of the changes in aggregate win rate are well described by the new meta.
The numbers don’t need to be pristine stats. We aren’t trying to write a peer reviewed scientific journal here.
For the purposes of the post, it’s pretty easy to say that the win rates observed in high legend for some decks implies a significant impact of skill, possibly even the vast majority.
But yes, the exact number is BS. That’s not known with any level of confidence.
1 Like
So after all this, you agree with me, because this is exactly what I’ve been saying since the start. Thanks for finally being honest.
IF you can. IF.
And you haven’t, not really.
The point is that he does not want.
Since the pure paladin nerfs NeonGhost kinda sees this entire talk about skill in hearthstone as some sort of war.
He isn’t trying to aggregate anything.
Just to win at any cost.
This is the why of this entire circus and personal acusations he does.
Oh so we’re in agreement then.
I haven’t and don’t claim to measure skill. I am calculating winrate per unit skill, where the unit of skill is defined as the average difference in skill between D4-1 and T1KL players that you take as “axiomatic.”
Winrate = matchups × popularity. Somebody tell him already.
Replying en masse to someone you have on ignore should be a bannable offense.
Who is “he” here?
Because the data we have doesn’t allow for it.
Nope, there’s nothing to win.
Nothing I say or do will make the op correct or less dumb.
Neon.
Gonna edit the POST properly it’s really ambiguous.
1 Like
I need to quote this again, because it’s 100% admission of the point I’ve made from the start.
This is why you never talk to the cops, people. An interrogation room is the worst place you can be. They have infinite time, they only need a single sound bite to convict you with, and they’ll honestly believe that they’re representing the truth as they twist your own words against you. And you will screw up eventually. You’re only human.
If this convo was just me and Neon I’d have dropped out a long time ago. I’m in this for y’all.
It’s still not entirely wrong. It’s just almost certainly not that exact number.
I just have zero idea what confidence interval it’s in. It’s likely low, but that doesn’t mean the number is meaningless, it just means it’s not exactly that number.
1 Like
Hard agree. Always remember, invoke your right to remain silent, even if instinctively you feel impolite in doing so. All it takes is an error in timeline, or whatever have you, to label you a liar to a judge or jury. And get a lawyer. Even with a public defender, albeit over-worked, you’re far better off than playing as your own lawyer, nor should you ever listen to jail lawyers. They’re in jail for a reason. My recent experience with jail has opened my eyes to the fact that these criminals want to give a) legal advice and or b) gain knowledge of your case for their own benefit (like snitching or extortion).
Anyway, sorry to derail here. Kindly disregard.
Then what good is it?
If you can’t separate what part is skill and what part is something else, then how can it be meaningful as a measure?
“Well, based on my calculations, baking this batch of cookies should have less than 12 cups of sugar for them to cook properly.” No idea how much it really takes or what else goes in the mix, but we know that cookies have sugar, so this is a valid statistic!
Also, you still need to link for me where VS does “exactly this” so I can read their work, because I know they would NEVER try to manipulate their data like this op does… they’re actually smart enough to understand why it’s useless and misleading. You can put ground effects on a mule, it doesn’t make it a car.
That’s what everyone should have done with the very first post, tbh.
To be fair I skim or skip a lot of your posts because they’re long and laced with insults… So I’m not intentionally misrepresenting you, I’m responding to the bits that catch my eye, like this
Apologies if I have misrepresented you.
This isn’t true. I gave my feedback, he dismissed my feedback. /shrug… It doesn’t upset me if people do some analysis that isn’t perfect. Why would it even hurt Scrotie’s feelings if I think his sample isn’t sufficient? I live and work in a world where people critique research and analysis day in day out, that’s science. It’s not offensive. We don’t put each other on ignore because we see flaws in a method. You dismissed my feedback about hitting people with your giant ego, and I’m not going to waste time arguing about that either 
It gives a vague idea of what decks you need to be better skilled at in order to succeed with, and which decks you’d want to drop at some point along the climb.
The number doesn’t need to be super precise for that.
Like when naga mage recent got re-nerfed. That deck was only playable in high legend skill levels. This kind of analysis highlighted that really well. The deck went from loses to basically everything to loses to almost nothing at the highest piloting skills.
We don’t need to know exactly what % of the win rate was skill vs meta vs other vague things, but it was still possible to identify that the deck had a massive skill component to its success.
The numbers don’t need to fall into a binary useless or tells all. This gives enough of a picture to make out what’s going on, even if it’s blurry and you’d prefer to have corrective lenses to see the small details.
False. It does not do this in any way.
But it does need to actually have any meaning at all beyond, “well, you need sugar to bake a cookie.”
No, this is still your assumption for the difference. Nothing about this or any other analsyis you’ve provided proves, quantitatively, that skill is the difference.
When you can’t reliably understand what factors are causing the difference you see, then the number is useless. It is, in fact, binary in this way. Garbage stat is garbage.
If you think you have something, go ahead. You’re wrong, but it seems like you guys would rather appear like you know something than actually know it.
So you accused people of things without actually taking the time to make sure you’re correct? That seems pretty rude.
And you and I both know the key word in that sentence is “reliable” and we both know why. That doesn’t say that there is no skill component, it says you can’t tease it out in a meaningful or reliable way from these numbers.
I am not bothered by it either.
The difference is that when your colleagues give you accurate and correct feedback, you accept their expertise because they actually have expertise…
I dismissed it because it was a falsehood or did you also skim past the part where the op indicated that his whole post was an attack? That my point about his motivations was confirmed? He literally told you he was out to get me, not the other way around, but go ahead and make me the villain - I don’t care.
He’s on ignore because he has proven to be a horse’s hind end who talks excessively on topics he doesn’t actually understand while condescending nearly everyone as a matter of habit.
I mean, last I checked he doesn’t even actually play hearthstone, but I assume over the last months he’s been happy to tell everyone why they’re wrong about everything without any real basis in experience or fact.
I mean, the VS analysis typically agrees with what’s been said here when it comes to skill testing decks. That’s half of the reason that I listen to the podcasts they do. It’s tedious to do this kind of calculation weekly for the decks in the meta.
It’s a lot of what drives players to choose different decks as they climb. Piloting differences across the board start to have a major impact on a deck’s performance as you escape diamond ranks.
Sure,you can’t perfectly isolate the skill part of these numbers, but you really don’t have to in order to make deck choice decisions, or to recognize that decks like arcane hunter just don’t hold up as well as you climb, even when the meta isn’t becoming less favorable compared to what it was in diamond.