Let's talk mastery curves. Gonna show some data and highlight some points on each, so let's go!

Steepest mastery curves:

Here are the top 8 "steepest" curves for games 1 through 100. Steepness is a measure of the difference between the game 1 WR and game X (100 in this case).

Inserted graph

Are these the mathematically HARDEST champions in League? Not necessarily. We're only looking at games 1-100. At 100 games, Azir's winrate is 51% while Nidalee (not pictured) is even LOWER at only 49% winrate.

Just because they don't have as large of a winrate gain in games 1-100 (steep curves), Nid (and Azir) still gain winrate well into the hundreds of games with additional mastery. We call these extended curves, you can see champions like Riven and Yasuo with them as well.

Compare these to flat curve champs. Common themes here are reliable CC and simple gameplay patterns. You can still get better/improve with these champs, just that your WR increase is more in playing League better overall, not necessarily from champ mastery (sponsored by @LSXYZ9)

Inserted Graph

This for my boys @Wickdlol & @IreliaCarriesU The Irelia update dropped her game fewer-games-WR, making her harder to plug and play, and I'll follow up on her when we have adequate sample for 400+ game analysis to see if her WR ceiling raises with her new mechanical demands.

Загрузка...

Inserted Graph

You've heard me constantly mention sample and getting adequate data. Here's why that's important: even looking at data from ALL time and across ALL elos, some champions just don't have the sample at higher # of games. How many players have exactly 389 Ivern games?? (not many).

Inserted Graph

Finally, to address a concern of @Hashinshin , let's address a hypothetical situation from Monday's mastery curves:

Let's say Riven has these winrates for game 1, game 400, and average: 37, 54, and 52

And Jax: 47, 54, and 52

How can they both have the same average winrate?

It's like saying 2+2=4 and then 2+1=4 right? Not quite. We have to consider weighed averages.

Riven might have a higher relative # of players with 100 games, and has more "54" data points. Meanwhile Jax might have a lot more game 1-10 players with 47-48% winrate.

More NOTE: to clear the air on sampling, I'll reiterate Monday's data:

Date: Jan 2017 thru Dec 2018 ELO: Mid Plat+

For all data in the mastery curves I showed today I used: Date: Jan 2017 thru Dec 2018 (except Irelia graphs for update timing, and Yas and Riven for sample) ELO: All