20 Second Timeout is the place to find the best analysis and commentary about the NBA.

Tuesday, March 09, 2010

Deflating Inflated Assist Totals

I only saw the tail end of New Orleans' 135-131 win versus Golden State, but during a late timeout Golden State's play by play announcer Bob Fitzgerald--noting that New Orleans' rookie point guard Darren Collison had been officially credited with 20 assists--said, "I'm going to say this politely but they need to watch the video of this game, because Darren Collison has about 13 assists." Jim Barnett, who is one of the best NBA color commentators in the business, agreed with Fitzgerald and cited a specific play in which a recipient of a Collison pass took "four or five dribbles" before scoring. Barnett correctly noted that an assist is not supposed to be handed out on such a play. Fitzgerald added that the New Orleans scorekeeper seemed to be "unclear about the rules" and that "this is something for the league to look at eventually." Barnett and Fitzgerald are not "homers" trying to belittle the opposing team--just seconds earlier, Barnett praised Collison's ability to get anywhere on the court off of the dribble, a trait that Barnett said Collison shares with top point guards like Chris Paul and Stephen Curry; Barnett and Fitzgerald are just confirming what I have documented on many occasions concerning the inflation of NBA assist totals, particularly regarding Paul (and now, apparently, this largesse is also being extended to Collison, who has replaced the injured Paul in New Orleans' starting lineup).

Collison's bloated assist total enabled him to break Paul's team record for assists in one game by a rookie and also tied Steve Nash for most assists in an NBA game this season. Fitzgerald is correct that the NBA should look at Collison's assists from this game. The NBA league office has more than once declared that box score statistics are reviewed and are subject to correction if any discrepancies are found; as a result of this review process, two triple doubles have been taken away from LeBron James, one because of an incorrectly credited rebound and one because of a bogus assist--yet the NBA has still not corrected some of the egregious errors that I have documented involving Chris Paul's assist totals, thus enabling Paul to set various team and league records in that category while also inflating his value in the systems used by "stat gurus" and undoubtedly increasing the MVP consideration that Paul received in previous seasons. Let me be perfectly clear: Chris Paul was the best point guard in the NBA in 2007-08 and 2008-09. However, there is also good reason to believe that his assist totals have been significantly inflated by overly generous scorekeeping. I don't know if this is because of bias, incompetence or a general loosening of scorekeeping standards but, whatever the reason, something should be done to correct such errors and to improve the overall accuracy of the scorekeeping process.

At a time when some writers are breaking their arms patting themselves on the back about supposedly being on the cutting edge of reporting about the alleged advanced statistics revolution it sure would be nice if the NBA and its scorekeepers took some steps to ensure that the most basic box score statistics are reasonably accurate; if we cannot trust the basic data then why should we believe in so-called advanced metrics that are 100% dependent on said data?

Labels: , , ,

posted by David Friedman @ 1:22 AM

6 comments

6 Comments:

At Tuesday, March 09, 2010 3:46:00 AM, Anonymous yogi said...

very well said. Reliable data collection is key.

however, keep in mind that some teams collect their own data - i'm sure they do a better job and therefore make better decisions it with their data.

 
At Tuesday, March 09, 2010 3:53:00 AM, Blogger David Friedman said...

Yogi:

I am sure that teams have access to better data than the general public. My point is that the player ratings devised by various "stat gurus" and promoted by so many bloggers are incomplete/inaccurate for a variety of reasons--not the least of which is that the basic data (not just assists but also turnovers, steals, blocked shots and other numbers) is unreliable to some extent. Rick Barry once told me that the only number he really trusts is free throw percentage, because virtually every other stat can be misleading either because of inaccurate recording or for other reasons (for instance, field goal percentage by itself says nothing about a player's range or his ability to create his own shot).

 
At Tuesday, March 09, 2010 11:58:00 AM, Blogger Bhel Atlantic said...

Given the problems you have noted with both Paul and Collison, perhaps the New Orleans official scorekeeper is more generous than other teams' scorekeepers? As I have noted before, if assist inflation is equally severe across all arenas, then it doesn't distort our ability to compare across players. (However, if assist inflation has worsened over time, then it distorts our ability to compare across eras.)

 
At Tuesday, March 09, 2010 4:12:00 PM, Blogger David Friedman said...

Bhel Atlantic:

The only way to determine the extent of this problem is for someone to go back over the game tapes and chart every assist. Every time that I have charted assists I have found that assists have been awarded too generously; I have not observed a single instance in which an assist should have been awarded but was not. The overall ratio between assists and field goals made has gone up over the years even though the NBA relies more and more on the isolation game.

The bottom line is that I have documented many instances in which bogus assists were awarded and the NBA never went back and fixed those boxscores; this is very odd, because the NBA clearly has the power to do this, as proven by the two triple doubles that the league took away from LeBron James.

 
At Tuesday, March 09, 2010 5:41:00 PM, Anonymous Anonymous said...

"if we cannot trust the basic data then why should we believe in so-called advanced metrics that are 100% dependent on said data?"

The problem, IMO, is that the assist is the most subjective of all of the recorded stats. (Well, I think the calling of fouls is more subjective, but that is on the official, not the statistician).

Unless the NBA decides to change the definition of an assist, it will remain that way.

To a lesser degree there is also subjectivity in rebounds and blocked shots.

 
At Wednesday, March 10, 2010 5:15:00 AM, Blogger David Friedman said...

Anonymous:

As I mentioned in my previous posts about this issue, assists are actually well defined in the rule book but those rules are not being followed. Overall, though, you are correct that assists are more subjective than most other basketball statistics--but that is no excuse for the NBA to ignore obvious errors or for "stats gurus" to disregard this issue and pretend that it does not have any impact on their formulas and conclusions.

 

Post a Comment

<< Home