Talk:Elo rating system/@comment-89.27.17.95-20120213021423

The way of determining the K-factor in LoL is incorrect. Weighing the first 15 games should be applied only in premade games. In solo queue the K-factor should be weighed against the player's own rating rather than total number of games played.

The current way of determining the K-factor causes that the error margin in mid Elo is outrageously high (±15%). If you examine the ladder, you can see that in some cases a player with 40:35 win:loss ratio may have a 150 points lower rating than a player with 35:40 ratio.

This should be fixed for the next season. At present, the ratings under 2000 Elo are only suggestive.