Does having the better player in some roles matter more for winning?

November 19, 2026 4 answers
Asked by scrimSubstitute+Kettle

I saw an analysis of roughly 38,500 League of Legends games comparing win rates when one team had a higher-rated player in a given role. Does it show which roles have the most impact, or is there more to it?

Answers

DOSbeeper_3R

In the pooled results, the estimated win rate gain for a 100 LP advantage was 2.11 percentage points for mid, 2.06 for jungle, 2.04 for top, 1.91 for bot and 1.90 for support. That puts mid first, but the differences are pretty small.

The comparison holds the other roles equal in the analysis. It tells you how strongly a rating advantage is associated with winning in that sample, not how much impact every individual player has in every game.

Reply to this answer
raidRole+Trombone

The regional breakdown makes a single ranking a bit shaky. Jungle had the largest estimate in North America and Korea, while top was highest in Europe West and Brazil. Bot was lowest in several regions, but support was lowest in others.

Those regional samples were much smaller than the pooled sample. Without uncertainty estimates, it's hard to tell which differences reflect a real pattern and which could be noise.

Reply to this answer
rankedQueue.Offbeat

Player rating isn't a perfect measure of skill in a particular role. If someone is off role, for example, their overall rating may not capture how well they play that position. The result also depends on how games and rating bands were sampled.

So I wouldn't use this to conclude that a role is weak or needs buffs. It measures one specific thing: the relationship between a role's rating advantage and the team's chance of winning.

rogueTalentByMug:

That also means the numbers don't directly tell you how much a role contributes when both teams have equally rated players.

Reply to this answer
8bitWanderer

Be careful with explanations for the regional differences too. The figures alone don't show whether teams in one region play around bot more, or whether players there are more willing to cooperate. Those are possible questions to investigate, not conclusions from this table.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions