NY Times: Viewing the N.B.A. Through a Statistical Lens
Posted by Justin Kubatko on November 12, 2010
This season we will be providing weekly NBA content to The New York Times. Here is a link to our first column:
Keeping Score: Viewing the N.B.A. Through a Statistical Lens
This should appear in the print version as well.
November 12th, 2010 at 12:14 pm
Excellent!
November 12th, 2010 at 1:50 pm
That's terrific. Time to educate the masses.
Not surprisingly both comments on the article are coming from the "If Kobe's not on the list, it's wrong" camp.
November 12th, 2010 at 2:02 pm
Jason J. wrote:
Not surprisingly both comments on the article are coming from the "If Kobe's not on the list, it's wrong" camp.
Yes, that was predictable. I submitted a response, but I don't think I'll have the time nor the energy to stay on top of it. Any help that BBR readers can provide would be appreciated.
November 13th, 2010 at 8:56 pm
After mulling over the article for a while, I've got a couple questions/thoughts on Total Rebound %.
1) If I understand correctly, it doesn't not use the actual number of rebounds available while an individual player was on the floor, so much as it abstracts an estimate from the total rebounds available for the game (or is it season?) compared to the player's minutes and rebounds.
2) That being the case, has there been any thought given to the possibility of the overall number of rebounds available dramatically changing when certain players enter and leave the floor (think Shaq for most of his career, vs someone like Telfair)?
November 14th, 2010 at 12:19 am
Congrats!
November 14th, 2010 at 5:34 pm
Couldn't we figure out the available rebounds exactly for each guy by simply looking at the number of missed shots when he was on the floor? Play-by-play data should provide this, no?
November 15th, 2010 at 9:46 am
Yes, but we don't have an easily-obtainable source for PBP data, and even if we did calculate Reb% from the PBP, we wouldn't be able to use it for years prior to 2006-ish anyway. Being able to estimate Reb% going back to 1971 is important because it allows us to put current performances in an historical context.
Although Nick's idea for a Reb% tweak in #4 is intriguing. We can estimate the % of team shots the player takes while in the game, and we obviously know his rate of making/missing shots relative to his teammates, so I think we could adjust the estimated # of rebounds available based on those data points. I'll have to take a closer look at this.
November 15th, 2010 at 11:59 am
Neil - I don't see how you'll be able to do that without PBP info. What you want to know isn't whether a player played in the same game as Shaq or Telfair, but whether they actually shared court time with a Shaq or Telfair. If I understand you right.
November 15th, 2010 at 12:27 pm
I'm just referring to whether Shaq decreased the number of estimated rebounds available to himself when he was on the court via his own high FG% (or Telfair increased the amount available due to his low FG%). This wouldn't apply to teammates, because like you say, we can't tell who was on the floor with him or for how long.
But it could apply to his own stats -- in theory, you would calculate the non-Shaq 2000 Lakers' shooting percentage and weight that by (1 - Shaq's FGA%), weight Shaq's FG% by his FGA%, and estimate the Lakers' shooting % when he was on the floor. Combine that with the Lakers' opponents' FG%, and you could get a truer estimate of how many boards were available when he was in the game.
Then again, the TRB% formula isn't denominated in missed shots, but rather in team TRB + opponent TRB, so you'd have to find a relationship between the two to make it work. But the basic premise is that Shaq decreased the # of boards available to himself by making such a high % of his shots, while Telfair increased the # available by missing so many shots.
November 15th, 2010 at 2:57 pm
I hope you can link to each weekly column.
November 16th, 2010 at 3:40 pm
I think there are two important lessons to learn:
(1) Statistical analysis can be a useful tool that can improve fundamental analysis.
(2) Statistical analysis is not a substitute for fundamental analysis.
That's the core issue that we're dealing with. I've seen too many stat geeks fly off the handle when their statistical conclusions are challenged. Conversely, the old-school people too often scoff at the numbers and miss opportunities to enhance their analysis.
I mean, come on. We all know that Kobe is one of the best players in the NBA right now, and Miami isn't one of the elite teams in the NBA so far. I don't think even stat geeks would argue that point. On the other hand, advanced stats often show why you need the 7-foot stiffs with horrible statistics or the point guards that average less than 5 assists a game - they're often better than one would think based on traditional statistics.