I saw an argument that rank filtering and patch timing can inflate the displayed average. How would that happen, and what does it mean when comparing champion win rates?
Why is the average Emerald+ win rate on Lolalytics above 50%?
Answers
The proposed issue is lookahead bias. If recent games are grouped using a player's current rank rather than their rank when each game was played, wins that helped someone reach Emerald can be counted as Emerald games afterward.
That can push the average above 50% without proving Emerald players won at that rate against lower-ranked players. The same effect could grow over a patch as more players change ranks, though the observed pattern alone doesn't establish how much inflation it causes.
Reply to this answer
The balanced versus mismatched game distinction is useful here. If both teams have the same number of Emerald+ players, those players collectively get one win for every loss in that game. Only games with unequal numbers of qualifying players on the teams can move the aggregate away from 50%.
One sample in the critique found about 8.3% of Emerald+ player-games were mismatched and about a 51.6% overall win rate. That result is consistent with the displayed average being elevated, but it doesn't isolate the cause by itself. The rank assignment method and the composition of mismatched games both matter.
Reply to this answer
For champion comparisons, don't treat 50% as the automatic baseline for every filter. If the site's Emerald+ average is above 50%, a champion just over 50% might still be below that filtered average. Compare within the same rank, patch and other settings, and keep sample size in mind.
Reply to this answer
The patch timing pattern is suggestive, not a complete proof. The critique says rank averages move back toward 50% when a new patch starts and then diverge again. A cleaner test would compare the same games grouped by rank at game time and by rank at analysis time.
Reply to this answer
Add your answer
Answers are reviewed before publication.
speedrunMenuKey:
So a game played in Platinum could end up in the Emerald+ sample if the player promotes before the data is analyzed?