29 replies to this topic

[Red94]

New post: Some stats that hint at just how good James Harden could become
By: Justin Wehr

If you had asked me yesterday whether I think James Harden might reach the stratospheric level that LeBron James, Kevin Durant, and Chris Paul are currently playing at, I’d’ve probably just looked at you funny. But then I plotted his Game Score data over his entire career (just as I did last week for Dwight Howard), and now the idea doesn’t seem so wild.

(Reminder: Game Score is a statistic invented by John Hollinger and available from Basketball-Reference that aggregates all of a player's box-score statistics into a single performance metric.)

There are two really encouraging things about this chart:

1) The trend. Look at how dramatically upward that thing climbs! James Harden has been getting better his entire career and there’s no reason to expect he’ll stop now.
2) The heights he reached last season. For perspective, LeBron, Durant, and CP3 are collectively averaging a Game Score of 21.8 this season. Last season, Harden played at or above that level for exactly one-third of the year.

These data convince me that not only could Harden reach the stratospheric level of the league’s best few players, but that it’s actually pretty plausible given the trajectory of his career.

One concern for many Rockets fans is Harden’s defense. While it’s a valid concern, I think it’s more than a little overblown. I have been manually tallying good and bad defensive plays for every game I’ve watched this season, and Harden has been below average but not as bad as some of the Rockets’ guards. And a lot of his below averageness can be attributed to his remarkably atrocious first game against the Clippers, which was far and away the most bizarrely bad defensive performance I have personally witnessed. Take away that game and he has been close to an average defender.

A bigger concern for me is Harden’s intangibles. Is he a good leader? Is he even a good teammate? I have no idea how Harden interacts with his teammates in practice or on the plane, but from what I see during games, it doesn’t look good. How many times do you see Harden celebrating for things his teammates have done rather than for things he has done? How many times do you hear him complimenting his teammates in interviews when not asked directly? One of my favorite moments from this season was when LeBron angrily confronted Chalmers during a timeout and then 20 seconds later called over to him, looked him in the eye, and said, “My bad, I was wrong.” Is Harden the type of guy who can say, “My bad, I was wrong”?

The bottom line is that I’m optimistic Harden can become one of the league’s best few players, but I’m not optimistic that he can take the next step, which is to be the type of guy whose teammates want to chew through a wall for him.

[Rahat Huq]

any chance we can get a post on the manual tallies at some point?  that's intriguing to me.

[thejohnnygold]

Cool!  I've been waiting for this data to come out.  Obviously, getting Omer back on the court would be so very nice.  That is a ridiculous spread he has.

 

Rather than look at the net score, I focused on the liabilities (blames).  Our 4 best players there were Asik, Jones, Parsons, and Beverley.  Out of curiosity I sought out some lineup splits to see if the numbers reinforced the eye test.  Check it out:

 

First, can a sample size get any smaller?  In a combined 11 minutes of action those 4 did post solid defensive numbers while still producing on offense.  Also, at the 19th spot subbing Casspi for Parsons worked well for all of 3 minutes.  Yeah, it's not much...but it does support the tallies!

 

For comparison, I swapped Asik for Howard.  Here's the chart:

 

Take from it what you will.  Tonight's starting line-up is up there at #2...with a negative +/-.    Also, for all those who long for Jeremy to get his due respect and be a starter...please take note of that #1 spot up there...the one with Beverley in place of Lin with the starters.  Better on offense and defense.  Better for the team.

[TTNN]

Cool!  I've been waiting for this data to come out.  Obviously, getting Omer back on the court would be so very nice.  That is a ridiculous spread he has.

 

Rather than look at the net score, I focused on the liabilities (blames).  Our 4 best players there were Asik, Jones, Parsons, and Beverley.  Out of curiosity I sought out some lineup splits to see if the numbers reinforced the eye test.  Check it out:

 

First, can a sample size get any smaller?  In a combined 11 minutes of action those 4 did post solid defensive numbers while still producing on offense.  Also, at the 19th spot subbing Casspi for Parsons worked well for all of 3 minutes.  Yeah, it's not much...but it does support the tallies!

 

For comparison, I swapped Asik for Howard.  Here's the chart:

 

Take from it what you will.  Tonight's starting line-up is up there at #2...with a negative +/-.    Also, for all those who long for Jeremy to get his due respect and be a starter...please take note of that #1 spot up there...the one with Beverley in place of Lin with the starters.  Better on offense and defense.  Better for the team.

 

with Howard, line up #17 looks impressive, how about #7, we should start with Brooks without Harden.

[timetodienow1234567]

Looking at those numbers Lin does not play well with Jones at all.

[thejohnnygold]

with Howard, line up #17 looks impressive, how about #7, we should start with Brooks without Harden.

 

Regarding #17, given the small sample size I'd bet the three ball was falling at a very high rate during that span.  While the Net +/- looks good, the defense isn't that impressive.

 

Regarding #7, I can only presume you're having a laugh.

 

A lot of those offensive numbers point to the idea of variance that Morey seems to be fond of.

[Richards]

According to your chart, Brooks looks much better than Lin and Bev with most line up if we are counting low minutes together. Doesn't seem right to me.

[Richards]

Oh again, we see what we wanted to see. Or numbers are deceiving us. 

Justin Wehr table says Lin > Bev, Asik > Dwight, Harden > Garcia. on defense.

[BrentYen]

I have a question, if we do not know what coaches actually told players to do defensively. How do you know it was a defensive blame (or a credit)?

[j_wehr]

Brent, I'm not sure I understand your question. Are you saying it's impossible to evaluate players without knowing the coaching assignments?

Thanks for the lineup stats, JG. As you noted, those data are very unreliable with such small sample sizes.

[timetodienow1234567]

I believe he was asking unless we know what the coach tells the players in regards to going under or over screens, hedging, doubling in the post and the multitude of scenarios that come up in the game , it's a little difficult to know if the player made a mistake or just did what the coach told him to.

[NorEastern]

A very concise post. Actually excellent information, a view that I had not thought about. Great job! It completely works with xRAPM and ASPM. IMHO Harden's overall defense only fails the eye test. I can find no statistical measure that says that he is a below average defender for a SG.

[thenit]

JG

 

THe lineup list is flawed I believe, not defending any particular player etc. Its so hard to judge our team this year because of the injuries. I think if we had a solid rotation of 70 games there could be indication on what line up worked or did not work. Right now due to injuries the sample sizes are so small that a great 10 min with one particular line up in 2-3 games against a weak team could enhance the stats and the opposite would be true if another lineup got killed for a big run against a tough opposing lineup. its so hard to value defence in the NBA that you always need as large as sample size as possible to judge a lineup or a single player through stats.

 

Exaple of all the regular rotation players Harden has the highest net negative and mistakes according to Justin but if we take out the clippers game his stats probably be the worst but not by a wide margin.

[j_wehr]

For those curious/confused about what the credits and blames mean and how I record them, this post has a little more detail and some older data.

[thejohnnygold]

I agree--the sample size is way too small here.  Our largest sample equals roughly 5 full games' worth of minutes.  Not very much.  I just thought it was interesting and agree that not much can be taken away from it.

 

I just wanted to see if the numbers supported the tallies--and I'd say for the most part they do.

[MrLobble]

What are the valuation metrics driving these analyses? Briefly looking at the analytical summaries, I would say these numbers are calculated on very broad assumptions that could actually be illustrating inconsistencies from our offensive opponents.

 

To discount the numbers correctly, I think you'd really have to break down each defensive possession, and include items such as direct penetration, bad rotation, missed boxout etc., to thoroughly identify what all causes our terrible defensive to break down.... (I assume this is what Morey's team does)

[HazeWinkle]

Stats are useless for this team because we already know they can score tell me something idk. I know they are bad at all 3 phases of defense and they cant buy a defense board in the 4th. Hell rockets are medicre a 6th or 7th seed at best   

Edited by HazeWinkle, 09 January 2014 - 02:16 AM.

[bboley24]

Seriously... who is this guy?

[Jatman20]

1) Some Rockets start running down the court to get out on fast breaks before we gather the defensive rebounds (an offensive philosophy). Some times we crash the boards well; but for the most part it has hurt the Rockets a great deal. 2) TO's (another by product of our offensive philosophy-up tempo style)
tonight we had about 9 or 10 and we won (avg 16 TO's/game). Grant it we played the Lakers. 3). Live by the sword-die by the sword. The Rockets fall in love with the 3 point shot (again part of the offensive philosophy (a numbers game....EX: if both teams shoot 45% from the field; the team that shoots 90 shots on goal will beat the team that shoots 80. In theory.....but other factors play into it as well. Rebounds, TO's, Opp teams 3's, FT's.). Rockets miss too many three point shots. Missed 3's by us lead to long rebounds by Opp team; which leads to fast break against us. Two games come to mind....@Pacers & @Thunder......both games we kept shooting bricks from 3's......while they refrained from shooting the threes. They kept adding two point shot after two point shot. Classic story of tortoise and the hare. The hired guns of Casspi and Garcia need to step up and prove their worth.

Side note.......all those not named D12 need to shoot better from the FT line. To pick up the slack.

Edited by Jatman20, 09 January 2014 - 04:44 AM.

[BrentYen]

Brent, I'm not sure I understand your question. Are you saying it's impossible to evaluate players without knowing the coaching assignments?

Thanks for the lineup stats, JG. As you noted, those data are very unreliable with such small sample sizes.

HI not discounting your great work, simply was just asking.

timetodienow1234567 explained my question well.

Thanks!