Post by eric on May 13, 2021 17:43:48 GMT -5
i was interested in how much a player's rebounding would change if they were on a better or worse rebounding team so i took a look
the tricky part is that a team's points and rebounds are very highly correlated (R^2 of .70!) and we don't want a player's usage to change too much, so we want two teams with very close points but very different rebounds, the best combo i found from last year (in the version of last year where nothing changed after sim 1) was the 76ers and Cougars (126 to 125 points per game, 50 to 54 rebounds) and Bass and Sharrar (24 to 22 points per 36, 11 to 13 rebounds)
o.k., so then i simmed the season five times with them on their actual teams and again flipped, here's how they changed
Uros Slokar: 10.3 to 10.2 r/36, 22.8 to 21.5 p/36
Lakers: 50.6 to 50.3 r/g, 114.1 to 115.0 p/g
SLOKAR CARE NOT PARAMETERS OF PUNY CAPITALIST EXPERIMENT
SLOKAR PARTICIPATE FOR HONOR GLORIOUS MOTHERLAND
the control case shows that points are a little shaky but rebounds are right about on the money in this sample size
right away we see that the effect (+1.3 -1.8 average 1.5) is smaller than the 2 we'd expect if rebounds were perfectly fungible. let's look at the players
Sharrar to some extent takes defensive rebounds from his teammates (because the 76ers only increased 1.3 when their rebounder increased 2.6), and since they're worse he gets less taken from him by Sixers than Cougars (so his personal number goes up)
similarly Bass "loses" some rebounds the Cougars were gonna get anyway (because they only decreased 1.8 when their rebounder decreased 3.8), and he sees a much bigger loss from playing with worse teammates than Sharrar did. it is possible to go way above the 14 rebound number, for example an all 100s SF in an all 5s league gets 27 rpg, so i don't think it's that he's hitting a de facto rebounding cap. the thing to do would probably be to look at the orb/drb split of the players but we can't very easily and i've stopped caring
the tricky part is that a team's points and rebounds are very highly correlated (R^2 of .70!) and we don't want a player's usage to change too much, so we want two teams with very close points but very different rebounds, the best combo i found from last year (in the version of last year where nothing changed after sim 1) was the 76ers and Cougars (126 to 125 points per game, 50 to 54 rebounds) and Bass and Sharrar (24 to 22 points per 36, 11 to 13 rebounds)
o.k., so then i simmed the season five times with them on their actual teams and again flipped, here's how they changed
Uros Slokar: 10.3 to 10.2 r/36, 22.8 to 21.5 p/36
Lakers: 50.6 to 50.3 r/g, 114.1 to 115.0 p/g
SLOKAR CARE NOT PARAMETERS OF PUNY CAPITALIST EXPERIMENT
SLOKAR PARTICIPATE FOR HONOR GLORIOUS MOTHERLAND
the control case shows that points are a little shaky but rebounds are right about on the money in this sample size
50.1 - 76ers w/ Bass
51.4 - 76ers w/ Sharrar
54.4 - Cougars w/ Sharrar
52.6 - Cougars w/ Bassright away we see that the effect (+1.3 -1.8 average 1.5) is smaller than the 2 we'd expect if rebounds were perfectly fungible. let's look at the players
11.3 - Bass on 76ers
10.1 - Bass on Cougars
13.3 - Sharrar on Cougars
13.9 - Sharrar on 76ersSharrar to some extent takes defensive rebounds from his teammates (because the 76ers only increased 1.3 when their rebounder increased 2.6), and since they're worse he gets less taken from him by Sixers than Cougars (so his personal number goes up)
similarly Bass "loses" some rebounds the Cougars were gonna get anyway (because they only decreased 1.8 when their rebounder decreased 3.8), and he sees a much bigger loss from playing with worse teammates than Sharrar did. it is possible to go way above the 14 rebound number, for example an all 100s SF in an all 5s league gets 27 rpg, so i don't think it's that he's hitting a de facto rebounding cap. the thing to do would probably be to look at the orb/drb split of the players but we can't very easily and i've stopped caring
