There are at least two methods for calculating winrate, and your description sounds like the worse one. In the opening post I suggest that the reader research expected winrate.
… my dude, I’m talking about the data you’re starting with, not what you’re doing with it.
Even the best jockey can’t win the derby on a mule, and the data is not suited for what you’re doing.
And you’re not understanding why it’s not.
1 Like
Why noone adressing my point about the other side of the equation.
If its nonsense then it shouldnt be difficult to explain why.
Not trying to be annoying or anything,i am just eager to learn why this wouldnt be a valid point.
Surely someone must have a good response but i am not seeing it yet. Even elchar doesnt adress the argument.
1 Like
All forms of win RATE exist in the latter category, not the former.
I mean for the love of Pete you had JUST said
and you don’t understand that winrate is fundamentally an analysis?
1 Like
That’s largely his argument.
You can’t say that the difference in win rates of decks at the top of the ladder is attributed to skill because maybe people are choosing to play other decks.
Like that doesn’t also apply to the rest of the ladder.
1 Like
I think NeonGhost has valid point.
The games from Legendary 1000 are very scarce in terms of a Mech Rogue player against a Enrage Warrior player who are willing to upload their plays. Only about 3% are Mech Rogue and 3% are Enrage Warrior players, that’s 30 players each. (well players switch up decks but let’s just say this for simplicity).
Considering most players don’t upload their games, those sample games could well be the same three Rogue player vs the same three Warrior players games, making those game results highly correlated and not independent enough comparing to the Diamond games.
In comparison, the Diamond 1-4 rank has 50 times or more population than the top 1k. I would say the Diamond 1-4 stats are better measure of the meta matchup winrate. It would be difficult to compare those two datasets.
3 Likes
Nope, it’s not.
It’s that you can’t quantify it with the data we have. You can’t make a statistic in the way this op does and say, well, here is skill, everyone.
I love how you keep coming back to I must not understand.
I regret taking you off ignore and trying to seriously explain the situation. It was clearly not worth the time to try because you can’t accept that you are just plainly wrong.
Good day.
But go one step further… you can’t even tell me if the excess wins at D4-1 were all from these same three players as they breezed through that D4-1 to get dumped in dumpster legend to just chill or if there were 50 other people who finished in top 1k and played rogue through the ladder grind then switched to a “fun” deck the rest of the month.
It’s not possible to separate these things out.
Or go two steps further and all the mech rogue players are in the 800-1000 range and how does that affect the other decks in the top 200 meta?
Again, it matters that our data set is not individual games by different players but a series of games by the same players.
Your point is good. For every point one side gets the other loses a point… making that side less skilled? It doesn’t work, you’re right.
I’m actually talking about why it doesn’t work and that’s because you aren’t measuring players you are measuring games. The skill isn’t the deck, it’s the pilots. You have to measure pilots to measure pilot skill.
Any indirect inferences are totally unreliable and theoretically suspect.
4 Likes
Stop doing this. There’s no reason to attack in such a personal matter.
Why won’t you stop?
1 Like
Which is why I chose a method that doesn’t rely upon that data at all. One way of looking at it is: the method I describe can isolate either the meta effect or the skill effect, and therefore I go for the meta effect first and derive the skill effect from that.
The raw data used in the calculations are D4-1 matchup winrates and top 1000 Legend deck popularity. T1KL matchup winrates are not used directly so there is no point there.
1 Like
Yeah, individual matchup data can fall short of the sample sizes needed to speak about that specific matchup with high confidence.
That said, across all of the top 1k legend games being recorded, mech rogue is seen a good amount, so it’s aggregate win rate isn’t just wrong because of these confounding factors.
Scrotie’s methodology here of comparing the lower rank matchup data to the meta breakdown at high legend (which there isn’t too much reason to doubt) isn’t bad.
There’s a disconnect between the two aggregate win rates. (How the deck does in diamond compared to high legend if you normalized the meta)
If the decks were being played the same way, there would not be a disconnect. I don’t know what else to call that difference other than skill differences.
The cause of those skill changes at the top legend isn’t really important. Even if the entirety of the loss in win rate is the actually good players arbitrarily changing to different decks, the deck they left is being beaten more because of skill differences in the deck pilots now using different things.
So if someone climbs to top 1k legend with pure paladin, then ditches it for something else that they now best pure paladin with… that’s still a skill difference at top legend, and the deck is going to be less good up there because people are piloting decks up there that beat pure paladin more than they might at lower brackets.
It’s still a skill difference. It’s not a direct measurement of skill, but you are still measuring an effect of it.
1 Like
I literally put them on ignore so I don’t have to see their pseudointellectual bull crap anymore… I need to do something more?
And why do you care so much? You haven’t done anything productive here,either. Best I can tell your sole purpose here is to attack my method of attacking their methods, which is pretty ironic, honestly.
Edit: For the record, I don’t think I replied to your point earlier, but I don’t think the game is rigged or the matching is rigged, I was pointing out the burden of proof and that many of the people handwaving assumptions don’t handwave assumptions when the topic is different.
5 Likes
As far as this thread is concerned, absolutely. If we concern ourselves with the cause of every cause, eventually we will trace back to the beginning of the universe.
On a purely theoretical level: idk maybe.
On a practical level: this will never happen. The vast majority of diamond players do not switch archetypes when they get to top 1000 Legend. They do not maintain the same deck archetype when they get to T1KL. They simply never get to T1KL because they can’t.
1 Like
Reading the rest of my posts would inform you that I have requested that you put forth an alternative hyptothesis or method of analysis. Instead you insult me, too.
1 Like
Neon doesn’t want to posit a hypothesis on virtue measurability because he doesn’t want virtue to be measurable. The only thing neon wants quantified are vices. Vices are what you point to to demonstrate your victimhood. But if you take virtues out of Schrodinger’s box then that whole scam falls apart.
1 Like
You understand that you edited that part into at least one if not more of those posts? That your initial, unedited posts were categorically hostile?
I’m not sure where you are getting an insult out of a factual accounting of what you’ve put here in this thread, which was to literally attack my method of attacking their method across multiple posts. Is that not what you’ve done here?
Look, I don’t care. It’s done. I just find it interesting that you don’t want to see that your role wasn’t magnanimous or, honestly, helpful.
4 Likes
A hearty chuckle was enjoyed reading this.
1 Like
I skimmed through the majority of the thread so I dont know if its been touched
I think you said that you tried to isolate things by mitigating deck popularity… but to me the whole skill difference is just the change in favorable vs unfavorable matchups at different ranks
Priest is the top dog at 2.34% per your table… but priest really only has favorable matchups against mech rogue / enrage warrior… (green on the heatmap) and in d5-1 mech rogue and enrage warrior are about 3% popularity… in top 1k legend where I think you took your data from that popularity explodes to 13% for enrage warrior & 8% for mechrogue… I’d argue your table shows the exact opposite of your closing statement. The 2% winrate would be explained by the fact that it has a favorable matchup against the 2 decks that explode in popularity at that level.
1 Like
The gap narrows, and the measurement of that narrowing is what counts. It’s an absolute value so it’s the same value whichever side of the table you’re on.
The challenge with your posts is that they have their own low signal to noise ratio.
It looks to an outsider like you think something is wrong, can’t quite explain it, and express your frustration instead. You obscure and diminish your own points by bookending them with pointless comparisons between your own intellectual “superiority” and the people you’re trying to converse with.
This is noise that has to be filtered out if people want to understand your points.
-what gap is narrowing?
“Absolute value”
Thats actually a good answer i have to admit.