Adjusted Plus-Minus
Posted: Thu Jan 6, 2011 5:16 pm
I get trashed for my obsession with obscure stats, and Adjusted Plus-Minus is my favorite. No one else seems to like it (or at least no one on this board uses it as much as I do). While I do think it is extremely under-appreciated, I ABSOLUTELY DO NOT believe that it is some kind of holy grail of stats that can tell us everything. I think it has to be carefully sifted through.
Why (I assume) people don't like Adjusted Plus-Minus:
1) It's new. It hasn't gained enough traction with longtime basketball fans, because data hasn't been available until very recently.
2) It's incredibly abstract.
3) It doesn't correlate with fantasy leagues at all.
4) It completely ignores traditional stats.
5) It has nothing to say after a single game. If people want to use stats to argue about why we lost to the Lakers last night, they have to go to the box score.
6) It doesn't give clear answers. There are always more layers, extrapolation, sifting to find meaning.
Why I like adj +/-:
Basketball is incredibly complex, and the fan experience is incredibly superficial. 90% of what fans know about basketball comes from highlight reels and box scores. Box scores teach us to value the most superficial aspects of basketball, stats that--without exception--can be amassed selfishly in ways that harm the team. The most obvious example of this is points--if a player scores a lot of points, he will be praised for the most glorified of box score stats, but he may in fact have hurt his team if he scored those points selfishly/inefficiently. But the same is true of every single box score stat (although obviously rarer with assists, fg%, and offensive rebounds, for example).
The most glaring problem with the box score is that it is an absolutely worthless gauge of defense. Measuring defensive value by steals, blocks, and rebounds is like measuring the beauty of a woman by leg length, freckle count, and breast size.
Adjusted +/- measures on-court subtleties like unselfishness, hustle, defensive focus, offensive spacing, chemistry, help defense, feel of the game, etc. No other stat comes close. Good basketball is built on these subtleties.
OK. So what?
I still believe in this team because I believe in the personnel. Yeah, we may be one or two players short from being a great team, but I think we can become a very good team given time to develop chemistry.
But I think we have to massively redistribute playing time. I believe that there has been enough time in this season to make some clear pronouncements about this team using adjusted +/-. The greatest weakness with adj +/- is that it needs lots of data to really be effective. It's still early in the season, but combining this year's data, last year's data, and old multi-season adjusted +/- gives us a lot to work with.
What we know for sure based on adjusted +/-:
1) We are still better with Steve Nash on the court. And it's not even close.
2) We are better playing four out. Sometimes even with seemingly awkward, nontraditional lineups (like Dudley playing the four).
3) We are better with four three point shooters on the court.
4) We are better with Channing Frye on the court. All six of this year's positive units have Frye. 7/10 top units last year in the playoffs had Frye (more than Nash, Stat, JRich, etc). 7/10 top units during the regular season last year had Frye (also more than Nash, Stat, JRich, etc).
5) We are better with Jared Dudley on the court. The top seven units from the regular season last year, the top nine units from the playoffs last year, and the top four units from this year all have Dudley.
What we can probably conclude based on adjusted +/-:
1) Amare Stoudemire is overrated. True, we didn't adequately replace him, but we are actually missing Lou more than Stat.
2) Vince Carter is massively underrated. He is one of the best players in the league, in part because he is one of only a few players who helps his team significantly on both offense and defense.
3) Josh Childress should play. He has the third highest adj +/- so far behind Nash and Carter.
Why (I assume) people don't like Adjusted Plus-Minus:
1) It's new. It hasn't gained enough traction with longtime basketball fans, because data hasn't been available until very recently.
2) It's incredibly abstract.
3) It doesn't correlate with fantasy leagues at all.
4) It completely ignores traditional stats.
5) It has nothing to say after a single game. If people want to use stats to argue about why we lost to the Lakers last night, they have to go to the box score.
6) It doesn't give clear answers. There are always more layers, extrapolation, sifting to find meaning.
Why I like adj +/-:
Basketball is incredibly complex, and the fan experience is incredibly superficial. 90% of what fans know about basketball comes from highlight reels and box scores. Box scores teach us to value the most superficial aspects of basketball, stats that--without exception--can be amassed selfishly in ways that harm the team. The most obvious example of this is points--if a player scores a lot of points, he will be praised for the most glorified of box score stats, but he may in fact have hurt his team if he scored those points selfishly/inefficiently. But the same is true of every single box score stat (although obviously rarer with assists, fg%, and offensive rebounds, for example).
The most glaring problem with the box score is that it is an absolutely worthless gauge of defense. Measuring defensive value by steals, blocks, and rebounds is like measuring the beauty of a woman by leg length, freckle count, and breast size.
Adjusted +/- measures on-court subtleties like unselfishness, hustle, defensive focus, offensive spacing, chemistry, help defense, feel of the game, etc. No other stat comes close. Good basketball is built on these subtleties.
OK. So what?
I still believe in this team because I believe in the personnel. Yeah, we may be one or two players short from being a great team, but I think we can become a very good team given time to develop chemistry.
But I think we have to massively redistribute playing time. I believe that there has been enough time in this season to make some clear pronouncements about this team using adjusted +/-. The greatest weakness with adj +/- is that it needs lots of data to really be effective. It's still early in the season, but combining this year's data, last year's data, and old multi-season adjusted +/- gives us a lot to work with.
What we know for sure based on adjusted +/-:
1) We are still better with Steve Nash on the court. And it's not even close.
2) We are better playing four out. Sometimes even with seemingly awkward, nontraditional lineups (like Dudley playing the four).
3) We are better with four three point shooters on the court.
4) We are better with Channing Frye on the court. All six of this year's positive units have Frye. 7/10 top units last year in the playoffs had Frye (more than Nash, Stat, JRich, etc). 7/10 top units during the regular season last year had Frye (also more than Nash, Stat, JRich, etc).
5) We are better with Jared Dudley on the court. The top seven units from the regular season last year, the top nine units from the playoffs last year, and the top four units from this year all have Dudley.
What we can probably conclude based on adjusted +/-:
1) Amare Stoudemire is overrated. True, we didn't adequately replace him, but we are actually missing Lou more than Stat.
2) Vince Carter is massively underrated. He is one of the best players in the league, in part because he is one of only a few players who helps his team significantly on both offense and defense.
3) Josh Childress should play. He has the third highest adj +/- so far behind Nash and Carter.