his conclusion that Joel is a very underrated player is accurate, but without really getting too deep into his statistical methodology I'd say it's mostly "well intentioned, but flawed"
Isn't that just ironic? An article on the value of Przybilla, which speaks of Przybilla being overlooked, but the author himself overlooked the spelling of Joel's surname.
The article would be more splashy with a list of the NBA's best rebounders, using his method. He seems to have only checked Przybilla. Also, how does he get the 35 in
While I fundamentally agree with the guy, this whole discussion is rather humerous. For years, certain posters on BBF used to lecture me about how over-rated rebounding was. Why? Because Zach was a good rebounder - and they didn't *like* Zach - ergo any positive contribution he made had to be dismissed. Nice to know that the popularity of Joel and Greg has made rebounding "cool" again!
A long time ago I did an estimate of this. I was curious how many wins are created over the course of a season by adding 1 point to a team's point differential per game. So I took about 15 seasons of data where a single data point would be (Team Wins, Team Point Differential Per Game). I then put those into Excel and plotted to see if it looked linear, which it did. I then let Excel fit a line to the data using linear regression. The slope turned out to be about 2.3 so each additional point per game on average adds about 2.3 wins. This agrees approx with the 35 number, since 82 / 35 is about 2.3. The fact still remains -- if Joel grabs a def rebound in his "zone" then he's still taking rebounds away from a team-mate x% of the time (maybe more than 50% of the time) but the article assumes that every additional rebound creates a possession.
If Joel is responsible for 6, then Brandon and LMA should be responsible for 20 each. They say Andre will bring in at least 5 more wins, and with Greg's improvement that should be another 10. Brandon (20) + LMA (20) + Joel (6) + Andre (5) + Greg (10) = 61 wins already!
So you looked up the difference in the final score, averaged for all games in a season, for each team each season, for 15 years. That's 450 (15 x 30) points. You graphed that on the Y-axis and team wins on the X-axis. You say (x, y) is (Team Wins, Team Point Differential Per Game). Linear regression fitted the best line to the cloud of points, and the slope was 2.3. That means that the average team wins only 1 more game if its season average margin of victory increases by 2.3 points. Maybe you meant that (x, y) is (Team Point Differential Per Game, Team Wins). If so, the average team wins 2.3 more games if its season average margin of victory increases by 1 point. That makes more sense. Then, scoring 35 more points over the season adds 35/82 of a point to the average difference and thus 1 win for the season (1 = 35/82 av points x 2.3 wins per av point). Okay, I got logic to jive with your statement, "so each additional point per game on average adds about 2.3 wins. This agrees approx with the 35 number, since 82 / 35 is about 2.3." "The fact still remains -- if Joel grabs a def rebound in his "zone" then he's still taking rebounds away from a team-mate x% of the time (maybe more than 50% of the time) but the article assumes that every additional rebound creates a possession." Yes, there are other assumptions, such as that the 35 points occur at moments when they make a difference to the game outcome. The author knew the defects in his model, which must be why he limited it to just Joel without presenting a list of many rebounders. Interesting article, but not one that will last. Anyway, the writer got an article out of it, which is what you do when you're a professional writer. Being an amateur, I myself know nothing about writing or arguing just to fill up space. Thanks for explaining the number 35.
