I could not have done a better job explaining this situation/problem then basketball-reference.com reader Eric. His comments (reformatted to make it easier for you to read) are as follows:
1. Eric:Thanks! I wanted to clarify one thing:
"You comment on them showing Kobe at the PG and SF positions as a evidence of that the statistics are misleading. What you failed to notice was they they only show him in those match-ups for 1% of the time. Certainly Kobe could have been cross-matched for a few plays during the season."
My point was not so much that Kobe being at PG/SF is questionable - I have no doubt that Kobe spent significantly more than 1% of the time defending the opposing PG or SF depending on who they were. My point was that I'm not sure that's what the 82games statistics even mean.
How do they decide when Kobe is playing SF on offense OR defense? What if both teams are playing a three guard lineup? What if they're playing the 2005 Celtics and Kobe is involved in doubling Paul Pierce? There's too much I just don't know about what 82games is doing in this particular area to reasonably evaluate their methodology.
I love 82games for on/off court stuff specifically because it is so clear what they are measuring, and because it is so clear what additional information has to be taken account to get a good evaluation from those numbers. By-position PER just doesn't have the same crispness.
2. My response: Wow, what a great comment Eric!
I have yet to read the article and hope to get to it soon (especially after reading your comment) but I think your comment lends itself as evidence to another way the modern game has really changed basketball: move and replace offensive sets have really blurred the line about how we determine players' positions.
For instance, this season Rondo usually brings the ball up and runs the offensive set, but there are a good number of times where when the primary option is not there Pierce or Garnett gets the ball at the top of the key (although this happens much less often than it did in the past.)
What do you consider Pierce and Garnett in these circumstances? They are being defended by forwards, but technically they are playing the point as they are at the top of the key. I would love for you, Eric, to expand on your thoughts a little more and hopefully I will be able to get a better reference point when I am able to read the article in its entirety.
3. Eric Says:
Thanks! What really got me thinking about these kind of issues was a player mentioned in #91: Tim Duncan.
Tim Duncan is described as a power forward, a center, or even a PF/C. People often cite his playing next to David Robinson as evidence of being a power forward, but this only shuffles the issue: how can we say David Robinson is definitely a center if we can't say Tim Duncan is definitely a power forward?
The issue is muddiest between PF and C, but it's a pretty prevalent ptroblem. There are some statistical trends that hint at an objective way to define positions, some that we would expect like more blocks/rebounds/fouls and less assists/steals going from 1-5, and some that are a little more interesting like percentage of rebounds that are offensive, but all of these lead to the somewhat uncomfortable reality of having two "centers" on the court at once, or three "shooting guards".
There's also the question of what strategy, if any, the coach is running: a point guard in the triangle offense is not the same as a point guard in D'Antoni's Phoenix offense, or Sloan's offense, or what passed for Mike Brown's offense - of course, how could they be?
As a coach, the job is to get the most out of the talent you have. If Steve Nash has a significantly different skill set than Derek Fisher, it only makes sense that his role in the offense would be significantly different. As such, if you are the fellow assigned to guard Nash one night and Fisher the next, I don't think it is reasonable to simply average the two as "point guards" if they are not doing similar things.
All of this makes me uncertain that any defensive measurement from box scores or conventional play by play is doomed from the start to approximation and guesswork.
If I don't see who Kobe is guarding, it's a good guess to pick who is nominally the shooting guard on the other team (as opposed to the center, for instance), but it's only a guess, and if the other team appears to have two or three shooting guards on the court at once, I'm completely flummoxed.
On top of all this, what if Kobe (or any perimeter defender) was told he would have interior help, and the help was absent? Put the other way, say Kobe is playing with Hakeem or Dikembe, or some all-time shotblocking great. Should Kobe get credit if whoever he is guarding eschews driving? Should we go the DRtg route and average team defense over 5 people? Questions I do not have satisfying answers to.
I have provided you with valuable information and innovative content I took a long time to create. Can you do me the huge favor of getting free e-mail or RSS updates by subscribing?