1 hour ago, AlexW said:To put is simply, there are a lot of variables that go into someone making the top cut (skill, list, luck, matchups), but something went right for those lists and players. The more that you see a list having success the more you can at least say the list itself is strong. As you get further down, the variance in those categories gets wider to the point where only one of them is consistent (the list), so it gets very difficult, imo, to tell how much the list has to do with its own placement.
Thanks. I agree with you about the variables. I'm thinking though that a large amount of entries will help to alleviate that problem. There are around 10'000 games to look at from just this regional season. That should somewhat remove matchup luck and general luck as factors, at least for popular squads (I'll get to that point in the reply to @baranidlo ). Player skill remains, but that is again a point where lists below the cuts are really helpful. A list that has several very low placings, or low winrates, could be hard to use.
1 hour ago, baranidlo said:Thanks for putting the statistics and graphs together, they are enjoyable and very insightful into how the global metagame look like.
However it's a mistake to try and look at these numbers as something more than what they are: indicators what a) is popular, b) works well , c) is easy to pick up .
Those are two very important points to be made.
1) On the global metagame: I can throw all kinds of regions together and see what comes up everywhere. It depends again on the amount of filled out lists how specific I can be. It is absolutely trivial, in fact a single line of code, to limit the whole analysis to a region or timeframe or both. But it loses power if only 30/92 players enter their data. That region will simply have to be ignored for such more detailed analysis.
2) On what the data really shows: yes! And again: that can be answered at least a bit by having full tournaments. It is why such a graph has some importance:
It does show whether a pilot is better than expected - and the expectation is coming from the overall advancement.
It also shows whether a pilot is very popular - Quickdraw is 4x as popular as Vessery, but makes the cut less often, relatively speaking.
Such conclusions can only be drawn if we have enough non-cut data, which is why I'm making such a fuzz and keep repeating myself.
1 hour ago, baranidlo said:Lists which pass all these 3 checks are the ones which are topping the metawing charts, and thus what you can generally expect to be present at the tournaments.
And that is precisely why I don't trust metawing as much anymore. Too many people misinterpret what is displayed, and it can get very frustrating. Don't get me wrong, I admire the work and result, and it can be used to do awesome things. But it is too easy to use, and too easy to misunderstand.
1 hour ago, baranidlo said:That is not only wrong, but also contraproductive and reinforcing the echo chamber doom and gloom topics, which have taken over these forums.
And that is another reason why we need the non-cut data. One motivation is to actually see how good certain things really are, or whether they are just popular. Harpoons are such a thing. GhostFenn is such a thing. It is the data below the cut that allows us to figure out that rebel lists are currently a bit stronger, as their relative amount increases from 49% to 54% from swiss to cut. But we also see that the shift is not that huge.
Edit: here is a simple win density plot for the Inquisitor, Lothal, Jake and Kylo in the TIE silencer
To me it shows that Jake is skill dependent but in the end just sucks - not one of the (only!) 9 players got more than 3 wins.
But that Kylo! That ship requires some skill! Peaks at both 2 and 5 wins are a very clear indicator for it.
I threw in Lothal Rebel and Inq for the fun of it.
Edited by GreenDragoonLothal, not Miranda
There's clear cases where metas aren't comparable - when the legal products differ, i.e. between different waves or when FAQs aren't yet in effect.
