About LoLWinCurve
Welcome to LolWinCurve, a champion power and difficulty analysis tool.
What is a win curve?
When players first pickup a champion, their winrate is low, but improvement is high. As they play more games and their winrate increases, the growth per game slows down. Eventually, the winrate stabilizes at some final value, beyond which additional games do not affect winrate. This is the "wincurve". While the general behavior is consistent across the entire roster, the exact shape of the wincurve varies - how low it starts, how high it goes, and how long it takes to get there depends on the champion.
Why does this matter?
The wincurve helps us measure champion power and champion difficulty.
In most matches, the players are still in the process of learning their champions. Therefore the average winrate is not reflective of the power of champions at their highest potential. The wincurve on the other hand presents this clearly - the "ceiling winrate" is the winrate at which the wincurve stabilizes.
Champion difficulty is a bit more complicated. What does it mean for a champion to be difficult? Is it when a champion is most punishing for beginners? Is it when the growth from floor to ceiling is the highest? Is it when the amount of games required to hit the ceiling is the highest? Is it when a champion is the strongest when played by the best? Regardless of the answer, all of these are observable from the wincurve.
What are "ceiling winrate" and "ceiling mastery"?
Ceiling winrate is the stabilization winrate of a champion - when additional games played no longer yield additional winrate. Ceiling mastery is the mastery value at which this stabilization occurs. LolWinCurve calculates these by taking a linear regression of the data, and, if the slope of the regression is positive (that is, more mastery grants more winrate), then the lowest mastery entries are tossed and a regression is taken again until the slope is no longer increasing. For champions where the sample size drops too low before the stabilization occurs, we return the highest winrate and mastery for which the sample size is adequate, and note that the true ceiling winrate and ceiling mastery must be above these numbers.
Why stop at diamond?
Assembling an accurate wincurve requires a lot of samples. There are too few masters+ games to make a wincurve for those ranks.