This uses a player's ten best offensive seasons, determined by offensive Win Shares. For players with less than ten career seasons, all of their seasons are included.
Rate stats are compared to the league averages (which are adjusted for the player's ballpark). For example, a player with a .290 batting average with a league average of .260 would be 11.5% better than league average. This number is then compared to the x players with the most plate appearances (where x = # of teams that year multiplied by 8), to find the percentile. This is done for all ten seasons, with the mean percentile (weighted on plate appearances) being the final value that is displayed on the graph.
As a general rule of thumb, high percentiles are good and low percentiles are bad. A batter in the 98th percentile of strikeout percentage is better than 98% of the league at avoiding strikeouts.
This uses a player's ten best pitching seasons, determined by pitching Win Shares. For players with less than ten career seasons, all of their seasons are included.
Rate stats are compared to the league averages (which are adjusted for the player's ballpark). For example, a player with a 2.50 ERA with a league average of 3.50 would be 40% (3.50 / 2.50) better than league average. This number is then compared to the x players in the league with the most batters faced, to find the percentile. This is done for all ten seasons, with the mean percentile (weighted on batters faced) being the final value that is displayed on the graph.
As a general rule of thumb, high percentiles are good and low percentiles are bad. A pitcher in the 87th percentile of walks allowed percentage is better than 87% of the league at avoiding walks.
This uses a player's ten best fielding seasons, determined by fielding Win Shares. For players with less than ten career seasons, all of their seasons are included. To qualify for a percentile rating, a player needs at least 200 career games at that position
Using Defensive Regression Analysis, this graph looks at runs saved per 1000 innings and compares this rate to the x players at that position with the most innings (where x = # of teams that season), to find the percentile. This is done for all ten seasons, with the mean percentile (weighted on innings) being the final value that is displayed on the graph.

