The higher level the hero is when the match ends, the more likely they are to win. We put the "all heroes" line there to show the curve so you can get a sense of how they compare to the "average hero" on a per-level basis and to show deviation from the curve.
thats not the point
in every match played every hero is between lv1 and lv25 = the stat includes all the heroes, and all the heroes on average have(or at least they should) 50% winrate
in ur graph every levels avg winrate is above 50% which leads to the conclusion that the average wirnate for all the heroes in all the matches is above 50%
how come?
I agree, makes no sense. Unless like you said ~1% of games are abandons and don't count as losses for the loser but still count as wins for the winner.
Also how the hell are like 20% of a heroes games finished by level 5? That's insane, I haven't had a game end under that level besides an abandon probably ever.
I'll give it another peek, but both of the things you're observing (higher than 50% win rate and high number of low-level games) are attributable to the (very high) rate of abandons and "no stats recorded" matches.
then perhaps it would make sense to treat abandons as losses at least for that particular stat - otherwise there is just no use for it
the number of low level games is surprisingly high, but it's at least possible between abandons, sand lp and botting
Does Dotabuff only record win rate on accounts that publicly show their games to dotabuff, the check mark option in settings?
发表评论请登录
there is a problem with either my brain or ur algoritms
Win Rate and Match Volume by Level: https://www.dotabuff.com/heroes/abaddon/trends
how can average winrate for all heroes on every single level be higher than 50%?
the only way i can see it happen is that you treat abandons not as losses but as something different, which, if some1 abandons in a losing team, techically results in 4 heroes losing but 5 heroes winning, in which case - fix it?