Glorious meta-data. Looks like either KF/UD is real bad, or people are drawn to playing bad versions of KF/UD, or bad players are drawn to playing KF/UD. I would guess the latter two. Other than that, those all look well within margins of player-skill variance. Thanks for posting this kind of stuff, Sok.
KF/UD is being dragged down by the following 3 players: Player A - 139 wins in 367 games (38% WR) Player B - 68 wins in 301 games (22.6% WR) Player C - 89 wins in 311 games (28.6% WR) I won't name and shame, but yea...
Yea, I noticed that was odd too, and isn't true, I am checking something now. EDIT: I mislabeled the games column in the charts above (they are actually WINS). WR is not affected tho. So in actuality they account for ~37% of the KF/UD games (2706), but accounting only for less than 30% of the wins. I am trying to find a good way of displaying the number of games, but the data structure is making it difficult to do it easily (tho it is easy to calculate the WR, ironically).
All fixed. Games is now properly games instead of wins. Note: If you were to add up the games total manually, this is about 180k games, compared to 92k as reported in the other thread because... each game has 2 factions players, so the number of "games by faction" will total to 2 times actual games played.
SO sok, @Skeezick Data suggests UD/FW (stitched) Needs some OP Juice Nah good stuff, good to see some of the data you probably work off.
IS has the lowest win rate for closely rated players and the third lowest win rate overall. So IS is.. not OP?
So whats up with ST having a much lower win rate with large rating differences? Just a lack of high elo ST players so the skill edge is always working against them or something? On a related note why does SP jump up so much, the rest all seem to be sort of close.
UD is most popular in FF as in splits (part in 4 of 8 most popular splits) 52% you face an opponent <100 points difference in ranking (looking at ff data only)
Huh, given individual Faction performance, I'd have thought FS/SL splits would have done better. it does seem like the 100+ Rating is what's killing it's performance, wonder what's goin' on with it. Anyway, as always, thanks for the data, it's interesting to look at and contemplate.