@Shyvadi0 like maybe, if midlane is 5 percentage points more popular than toplane, it shouldn't get aegis ~3x as often?
Like why should I queue top over mid, if queueing midlane just gives me +2lp per game on average
@iterogg champ needs be able to lane vs mf tahm
needs some sort of counterplay vs enemy wombo combo, so like either mobility or good defensive spell or absurd output. Champ should be good at dealing with J4 Galio lategame, as Diana and Graves wont Peel.
Nilah, Sera, Brand, Kogmaw?
@e_stratson@mrginden generally AI learns to extrapolate from data. Maybe it learns a vector that corresponds to 'Oil -> Pencil'
now if you apply that vector to something that is not Oil and you consider that a new style, technically the information for that vector was contained in training data
@drewlevin EUW, gameID 7880326918
Kaisa gets angry at leona and then starts to run it down on all lanes. afk in jgl or trying to bait her teammates into death while running it down.
kaisa name is blaquatre#EUW
@Scarfang_@drewlevin you do understand that the expected pre reset team lp difference between teams would be 1800.
an example that would fit exactly in the expectations is a lobby with 9 gold iv players and a master play.
and it probably would take 2 months to resolve.
@yanatsundere@zak_vvv last time I was tilted by ranged top I looked at it and it was 17 out of my last 40 games. mind you that was a biased sample, I only looked cause it felt so frequent. (the probability that enemy top is ranged is roughly 15-20%)
@Crucile while I would agree if I mostly cared about competitive integrity, I also feel like league is a game, and people should be allowed to play games for fun if it doesn't disrupt the overall experience too much
@shy_rony@Arktheos there is a reason to press blue: you care about someone who might have pressed blue.
also while 100% of ppl picking red has 0 casualties, so does 100% picking blue. This is an example where the optimal choice for the individual can be the suboptimal choice of the group
@Citriclol 1% of the players lose on average like 250 lp worth of mmr.
now these mmr are entirely lost in the void.
but if you distribute them through all players, that's 2.5lp worth of mmr per player throug the total playerbase.
Dia players will notice, besides it should be negligible
@zac_gaming__ I mean yeah, it's perfectly reasonable talon is a full item ahead of kayle because he is fed. But the general idea was that jgl has more access to gold and map impact while top has more access to exp.
@geoclidean f(1) - f(0) = f(1) - f(1-1/n) + f(1-1/n) - f(1-2/n) + f(1-2/n) - ... - f(0) > n sqrt(1/n) for all n.
the right hand side is unbounded, so such a function cannot exist.
@Pilgrim__Jay@ItakGol@FrankFelixAI Now, if all distributions change in nature if you sum them up except e^-x^2, then it becomes likely that everything converges to that fixed point.
(generally this is not the only fixed point, but it's the only fixed point with finite variance)
@Pilgrim__Jay@ItakGol@FrankFelixAI Two of the easiest functions that are symmetrical and taper to 0 as they go to infinity are:
1/(1+x^2) and e^-x^2
when summing two probability distributions, you use the convolution operator, which stays similar for e^-x^2 while it completely changes for 1/(1+x^2) and others
@GriffinHQ_@AryaGuGo oh true, meant to say 1 hour 25 mins
also that is what happens when one role has twice the player base as another role. if mid is 30% of players and sup is 15%, then each midlaner only gets into every 2nd game, which means he waits on average the duration of 1 match and 2 queues