Huh. They only represent 50% of the field making the cut here so its a step in the right direction. Still this is troublesome, if only there were a compiled list of top faction appearances. To keep things modern lets say from Nov. 2014 to now taken from an online store that keeps track of lists to sell you, say Discount Games Inc. This information would surely continue to correct these results:
Cryx - 10
Circle - 9
Skorne - 5
Cygnar - 4
Trollbloods 4
PoM - 3
Legion - 3
Ret -2
Minions - 1
CoC - 1
Well fuck. At this point I would be worried that the data is pointing towards some real unfortunate trends. Thankfully the competitive WM braintrust was right there to clear it all up:
"Players win games not factions"
"It's the player."
"Do remember both minions and ret have won major cons. It's all about the player."
The mental gymnastics on display when you've biased yourself to certain outcomes are incredible.
If I get a moment I'll try to scrounge up some more data; but finding it is a motherfucker.
ReplyDeleteAn interesting thing about the Skorne results: basically all of the placings above are Mord/Fist pairs.
In the IG brackets SOS is the real killer of the spread. You can't fix bad match ups. Look at the 3 win guys and you have a pretty decent spread.
ReplyDeleteMasters that data set is within the SD for such a small group.
Let the data tell you what you see, Mr mental gymnastics.
In conclusion, haters gonna hate playas gonna play if you don't like the game you don't have to play.
SoS, as in the people at the top of the 3 wins bracket did better against the tougher field? I'm not sure that is making the point you want it to.
ReplyDeleteI would love to see those standard deviation calculations.
It's not about hating so much as the importance of learning the meta. To be really competitive it's vital and WM meta theory is both lacking in data and is infantile. It's puzzling.
You have fun with your trash analytical skills
ReplyDeleteThat is rather mean spirited sir.
ReplyDeleteAlso as you didn't answer the standard deviation question, the answer is it isn't within a SD for that data set. It's much too small of a sample to get a reliable number from anyway.
This forum has gotten way more interesting.
ReplyDeleteI'd like to take a look at this problem eventually but there are tons of modeling assumptions/problems. Standard deviation doesn't make any sense in this context.
One way to treat the data, without doing something nice but complicated, is to treat the data as a Kx2xT table where K is the number of factions and T is the number of tournaments. Your outcome of interest is whether or not a faction made the top 25% (or something like that).
With this setup you can do some kind of Chi-test to test the odd's of winning with any particular faction. This setup also helps you deal with some of the confounding in the data and deals with the non-independent nature of the observations. It also has the right support; i.e. count data.
I'd like to do something nicer, but this seems reasonable and attainable.