← Research

How much champion mastery does it take to start winning?

Published September 30, 2026 · 39,039,573 ranked player-games from 2,867,755 players · NA, EUW and KR · September 11–22, 2026

What we measured

We grouped each ranked appearance by the mastery points that player had on that champion before the game began, then measured how often the player’s team won.

• 39,039,573 player-games from 2,867,755 players on patch 16.18, across North America, Europe West and Korea.
• Six mastery bands, from under 5,000 points to over 300,000, using the same definitions as KayLoL’s champion pages.
• Mastery, rank and the smurf-like-account check were known before game start. The game’s result was used only as the label.
• The 95% intervals resample players, not individual appearances, so repeated games by the same player do not pretend to be independent evidence.

The curve rises almost five points

The lowest-mastery group won 47.31%; the 300,000-plus group won 52.20%. That is a 4.89-percentage-point spread from one end of the curve to the other.

The curve reaches 50% inside the 50,000–150,000 band. Interpolating between the band centers puts the crossing at about 71.1k points.

The break-even point changes with rank

Bronze crosses first, at about 44,000 points. From there the threshold climbs steadily through Diamond and Master+. Iron is the exception: its 50,000–150,000 band sits almost exactly on 50%, so a small movement pushes the interpolated crossing much later.

Read these as population thresholds, not quotas. Reaching 141,000 points does not cause a Diamond player to become good, and a particular champion can have a very different curve.

Smurfs barely move the overall answer

Excluding seats whose pre-game rank-for-mastery score was more than three standard deviations high moved the all-ranks crossing from 71.1k to 72.3k: a shift of 1.2k, about the same size as the drift between the first and second half of the patch.

The effect is larger at high rank. The Diamond threshold moves about 10,000 points later and Master+ about 20,000, because low-mastery high-rank accounts win two to three points more often than ordinary accounts in the same mastery band.

The three regions tell the same story

The exact crossing differs by server, but removing smurf-like seats shifts every region by only about one to two thousand mastery points.

What should a player do with this?

Treat 71.1k as a description of the ranked population, not a deadline. The useful result is the shape: unfamiliar picks carry a real average cost, the largest gains come before 150,000 points, and the curve continues upward after that.

Your own answer depends on the champion. See the population curve and your personal pool on Champion Pool, or open the champions page to compare each champion’s first-time floor with its 300,000-plus result.

What this cannot prove

• This is not a causal learning curve. Nobody reaches 300,000 mastery points by accident: players who enjoy or succeed on a champion are more likely to keep playing it, while others stop.
• Pooling champions hides large differences in both starting strength and learning curve. The result also describes groups of players, not the path of one player followed from zero mastery.
• The frozen study covers 12 complete days of one patch. Its comparisons are like for like, but another patch can move champion balance and the exact crossing.
• Iron’s threshold is unstable because one whole band is statistically indistinguishable from 50%. Its point estimate should not be read more precisely than its interval.