APBRmetrics

After the Spurs' 3 entries, the '08 Lakers? Really?
Kobe was in top form. Gasol only was there for the last 27 games, and then 21 in the playoffs. Bynum was MIA (after 35 RS games).

Did you extrapolate Pau to 82 games to get his 'impact'? Or were you using playoff VORP?
If it's playoff VORP, are you measuring the composition of the teams? Or how well they performed in the playoffs?
Playoff VORP (or BPM, or anything) is of course depressed by the level of competition; and RS rates aren't necessarily applicable -- or need extrapolation.

Mike G wrote:

Code: Select all

Gini   
.253   sas2014
.311   sas2013
.400   Sas2005
.402   lal2008
...

After the Spurs' 3 entries, the '08 Lakers? Really?
Kobe was in top form. Gasol only was there for the last 27 games, and then 21 in the playoffs. Bynum was MIA (after 35 RS games).

Did you extrapolate Pau to 82 games to get his 'impact'? Or were you using playoff VORP?
If it's playoff VORP, are you measuring the composition of the teams? Or how well they performed in the playoffs?
Playoff VORP (or BPM, or anything) is of course depressed by the level of competition; and RS rates aren't necessarily applicable -- or need extrapolation.

No, I was using regular season VORP. The gini of (6 Kobe,4.3 Lamar ,1.6 Bynum,1.6 Gasol,1.2 Farmar,1.1 Turiaf,1.1 Radmanovic,1 Fisher,.8 Walton,.6 Vujacic) is .402 (at least according to the R package I'm using).

Compare that to the Spurs of '05: (5 Ginobili, 4.7 Duncan, 3 Parker, 2.1, Bowen, 1.9 Horry, 1.8 Barry, 1.5 Nesterovic, .7, Brown, Udrih .4, Rose .3). The Spurs had a much more equal top 7 players (.249 vs .363) but after that they have a drop off.

Your concerns of course beg the question: is it really suitable to use a Gini coefficient on this type of metric, and if it is how many players do we want to consider?

I'd think that if you're predicting postseason success, you only want to consider those players who are in the playoffs. Bynum's 1.6 are irrelevant, and Gasol's 1.6 over 27 games would be 4.8 over 81 games.

It's therefore more meaningful to use BPM, plus 2 or 3 to avoid negative values. If a player doesn't finish the season, that changes the team balance in the postseason.

Mike G wrote:I'd think that if you're predicting postseason success, you only want to consider those players who are in the playoffs. Bynum's 1.6 are irrelevant, and Gasol's 1.6 over 27 games would be 4.8 over 81 games.

It's therefore more meaningful to use BPM, plus 2 or 3 to avoid negative values. If a player doesn't finish the season, that changes the team balance in the postseason.

Alright I will look at BPM + 3 or so. When I have more time I'll check with postseason stats to get the players who actually played, but use their regular season BPM as an indication of skill? Also, I think I'll look at top 10 players in minutes not by BPM (or top 8 or whatever).

Mike G wrote:I'd think that if you're predicting postseason success, you only want to consider those players who are in the playoffs. Bynum's 1.6 are irrelevant, and Gasol's 1.6 over 27 games would be 4.8 over 81 games.

It's therefore more meaningful to use BPM, plus 2 or 3 to avoid negative values. If a player doesn't finish the season, that changes the team balance in the postseason.

but we don't care about predicting postseason success; we only care about understanding the distribution of impact among a team's roster that is good enough to win a championship.

I am not sure what could/should be done about players missing time due to injury but as BPM doesn't take into account playing time, it could have some quixotic results.

You probably want a measure like VORP/G, and only for those players in the rotation during playoffs?

permaximum wrote:.. I knew Pistons of 03/04 would demolish Lakers with Shaq, Kobe, Malone and Payton and I never even looked at one stat or metric to see that...

So, how did you know this would happen?

Mike G wrote:

permaximum wrote:.. I knew Pistons of 03/04 would demolish Lakers with Shaq, Kobe, Malone and Payton and I never even looked at one stat or metric to see that...

So, how did you know this would happen?

Chemistry. The pieces that Pistons team had 100% spot on. On the contrary, Lakers' game was all over the place. People are not machines you can't sum their positive effects on the court when you pair Shaq, Kobe, Malone and Payton together.

Luckily I watched a lot of Lakers' and Pistons' games that season, that's why I didn't need to check any stat or metric then.