The raw data behind the story "The Best MLB All-Star Teams Ever" https://fivethirtyeight.com/features/the-best-mlb-all-star-teams-ever/.

mlb_as_team_talent

Format

A data frame with 172 rows representing Major League Baseball seasons and 16 variables:

yearid

The season in question

gamenum

Order of All-Star Game for the season (in years w/ multiple ASGs; set to 0 when only 1 per year)

gameid

Game ID at Baseball-Reference.com

lgid

League of All-Star team

tm_off_talent

Total runs of offensive talent above average per game (36 plate appearances)

tm_def_talent

Total runs of fielding talent above average per game (36 plate appearances)

tm_pit_talent

Total runs of pitching talent above average per game (9 innings)

mlb_avg_rpg

MLB average runs scored/game that season

talent_rspg

Expected runs scored per game based on talent (MLB R/G + team OFF talent)

talent_rapg

Expected runs allowed per game based on talent (MLB R/G - team DEF talent- team PIT talent)

unadj_pyth

Unadjusted Pythagorean talent rating; PYTH =(RSPG^1.83)/(RSPG^1.83+RAPG^1.83)

timeline_adj

Estimate of relative league quality where 2015 MLB = 1.00

sos

Strength of schedule faced; adjusts an assumed .500 SOS downward based on timeline adjustment

adj_pyth

Adjusted Pythagorean record; =(SOS*unadj_Pyth)/((2*unadj_Pyth*SOS)-SOS-unadj_Pyth+1)

no_1_player

Best player according to combo of actual PA/IP and talent

no_2_player

2nd-best player according to combo of actual PA/IP and talent

Source

http://baseball-reference.com , http://chadwick-bureau.com, Fangraphs