James Harden is a superstar

[Doctor MJ]

can anyone explain why the Rockets have technically been better with Harden off the court all season? I didn't want to bring it up before because people would just yell sample size, but now that the season is over we can discuss it. His net rating all season has been between -1 and -6, and is currently -3.4. For comparison, Kendrick Perkins is +0.4

Asik is a net +7.3, so it's not just the starting unit. Delfino is even +5.9, so it's not just Asik.

I'm not saying that Harden makes the Rockets worse, so why do the numbers say he does?

I'm frustrated with the lack of advanced +/- data right now, so I'm kind of gunshy using any of this data at all really. (Funny how spoiled I've gotten by it all.)

Getting into it some though:

First thing to look at is minutes played. Guys who've played way less minutes oftentimes have major biases in their raw +/- data. The top 3 guys minutes-wise all have similar raw on/off stuff to Harden. The 4th guy on the list is Asik, who has the sweet net +/-, but plays about 9 minutes less per game than Harden. Okay so when I see something like that, if I'm asking who is more valuable, I'm pretty reluctant to pick the lesser minute guy. The guys playing even less minutes btw, just ignore them when looking at Harden. It's not realistic to think that they are doing something secretly amazing. One can argue that the drop off between Harden & them isn't as huge as people think, but not more than that.

Forgetting about the glory each guy deserves and just getting into the nitty gritty, the glaring thing I see is that the team's rebounding falls apart when Asik goes to the bench, and this contributes to the clear boost Asik seems to be giving on defense. Looking into the team's leading rebounders we quickly see what appears to be an issue:

1. Asik 11.7
2. Parsons 5.3

Okay, so, rebounding without Asik appears to be not getting done. That's a bit melodramatic because it's not so much that the team is horrible at rebounding without Asik as it is that they are just quite good with him, but still, Asik's rebounding is having a big enough impact that it's distorting all other players +/- on the team.

So yeah, I think the issue is that Asik is really good, and the team hasn't really found a way to work that well when he's not on the court.

Now one can ask: Shouldn't Harden, as the "superstar" be able to trump anything along those lines? I'd like to have access to more data, but I'm inclined to say, yeah, this is a chink in his armor. Harden does need to get in the flow a step further before he's really an MVP candidate type of player.

[The Infamous1]

He's another example of a guy who's PPG doesn't show how flawed he is as scorer. It's still just long volume 3's, Pick and roll and tons of free throws. He can't get by his man without a screen unless its a scrub thats why hes so heavily reliant on the pick and roll. Heck watching the lakers first quarter it's the same **** and this backcourt is Dleague level right now.

I'm not a big harden fan but fallacys numbers are perfect proof why plus minus numbers are absolute garbage.

[therealbig3]

Steve Nash is one of the best offensive players of all time, and he couldn't really get by his man one on one either...he almost always needed a screen.

He was better than Harden, but acting like getting a screen and operating out of the PnR is some sort of disadvantage is ridiculous...it's the most effective offensive play in basketball when executed properly.

[mopper8]

Since when is beating someone off the PnR easy? Again, like the point made in the 'Melo thread about finishing PnRs, if it were so easy, everyone would do it. It's not.

1 - you have to be good at reading the big man. Most guys aren't!
2 - you have to make snap decision about taking the screen or rejecting the screen
2b - if you reject the screen you need the handle to get by your man
3 - you need excellent timing on when to go--go too early, and you end up with an offensive foul for a moving screen; go too late and driving lane is cut off
4 - you need to be able to lead your man into the screen
5 - even if you can do all those things to get by your man, you still need the handle to get into and through the lane without getting stripped
6 - you need to be good at finishing in the lane, or have a good midrange game, or (preferably) both

That sets aside the shoot/drive decisions, the shoot/pass decisions, etc.
There's a reason why there are only a few guys in the league truly great on the PnR. The fact that Harden is one of them speaks highly of him.

[nunemouse]

I'm frustrated with the lack of advanced +/- data right now, so I'm kind of gunshy using any of this data at all really. (Funny how spoiled I've gotten by it all.)

http://talkingpracticeblog.com/2012/10/24/hello-world/

I think that does a pretty good job of creating an APM stat. You can also do more with the results than you can with RAPM, their breadth of application is larger.

In it, Harden has a 4.8 IPVo (4th best, behind Lebron, Paul and Durant) and a -1.0 defense (makes sense), giving him a net of 3.8 IPV (12th best in league).

[mopper8]

I will tell you what though, watching Hou-LA right now, Harden seems to love passing the ball at the end of the shot clock and letting someone else take the bail out shot, so I wonder how that effects his % at the margins.

[CKRT]

He hasn't really been doing it all season. He was really off tonight.


Edit:

What's going to be interesting is the narrative against Harden is going to be lol worthy. "Oh he led them to the 8th seed big deal" when the Rockets could of easily been the 6th seed.

Harden's numbers for this season:
26/6/5 on 44/37/85 on 17 shots.
23 PER, 60%TS, 50% eFG% 116 ORTG and 106 DRTG, and 3.8 TOVs.

Good season for him. A solid playoffs against OKC will really cement him as a top 5 player.

[G35]

He hasn't really been doing it all season. He was really off tonight.


Edit:

What's going to be interesting is the narrative against Harden is going to be lol worthy. "Oh he led them to the 8th seed big deal" when the Rockets could of easily been the 6th seed.

Harden's numbers for this season:
26/6/5 on 44/37/85 on 17 shots.
23 PER, 60%TS, 50% eFG% 116 ORTG and 106 DRTG, and 3.8 TOVs.

Good season for him. A solid playoffs against OKC will really cement him as a top 5 player.

Yeah like the narrative wouldn't have been just as ridiculous claiming Harden was a top 5 player if the Rockets would have gotten the 6th seed. Look at what Harden did in a playoff like atmosphere.....

[Peregrine01]

His numbers have improved tremendously since last season, mostly due to his increased role. However, his play has fallen off quite a bit since the all-star break. Often settling for bad shots and dribbling out the clock.

I don't think that he is top-10 status and I sincerely hope that he hasn't been buying into his own hype.

[CKRT]

He hasn't really been doing it all season. He was really off tonight.


Edit:

What's going to be interesting is the narrative against Harden is going to be lol worthy. "Oh he led them to the 8th seed big deal" when the Rockets could of easily been the 6th seed.

Harden's numbers for this season:
26/6/5 on 44/37/85 on 17 shots.
23 PER, 60%TS, 50% eFG% 116 ORTG and 106 DRTG, and 3.8 TOVs.

Good season for him. A solid playoffs against OKC will really cement him as a top 5 player.

Yeah like the narrative wouldn't have been just as ridiculous claiming Harden was a top 5 player if the Rockets would have gotten the 6th seed. Look at what Harden did in a playoff like atmosphere.....

Yeah, good thing Kobe succeeded in a playoff atmosphere in Utah.


Oh wait, he air balled shots like it was his job.


See, I can play the **** argument game too.

[G35]

Yeah, good thing Kobe succeeded in a playoff atmosphere in Utah.


Oh wait, he air balled shots like it was his job.


See, I can play the **** argument game too.

You can but you were playing for Harden. I told you that you can ALWAYS turn the narrative to the other side.

Also Kobe was a rookie....an 18/19 year old rookie....and no one had declared him a superstar yet. Not like how people are so sure that Harden is. And this was against an 8th seeded Lakers team that didn't even have their best player on the court. Utah went to the finals that year. Big difference.

See how we can go tit for tat. Why don't the Harden apologists take the L tonight and throttle back on all the superstar talk.....

[EArl]

Kobe was a rookie the year he air balled the shot. I don't know why you're so mad.
This is harden's first season coming off the bench. The playoffs will be a testament if Howard belongs in the Superstar group.

[Doctor MJ]

I'm frustrated with the lack of advanced +/- data right now, so I'm kind of gunshy using any of this data at all really. (Funny how spoiled I've gotten by it all.)

http://talkingpracticeblog.com/2012/10/24/hello-world/

I think that does a pretty good job of creating an APM stat. You can also do more with the results than you can with RAPM, their breadth of application is larger.

In it, Harden has a 4.8 IPVo (4th best, behind Lebron, Paul and Durant) and a -1.0 defense (makes sense), giving him a net of 3.8 IPV (12th best in league).

Maybe you can explain this more to me. So far as I can tell the stat uses SPM which means its absolutely not APM. Am I wrong? Fine if you like hybrid models, and I get why my post seemed to cry out for some knew nuanced stat, but I would never use a hybrid to try to get to the bottom of more raw +/- data as it clearly adds more mystery into the mix.

This is also a general philosophy. When you say you can do "more" with the results, to me that smacks of the kind of holy grail thinking we often see from statisticians. I'm sure that if I were forced to use only 1 stat, that 1 stat should be a hybrid...but I'm never going to use only 1 stat. If you want to really win me over, what I'd need is to feel like I could fit this in with my other non-+/- routines without adding serious unknowable redundancy issues. If I'm already using the box score to influence my +/- data, then how can I see what the +/- contribution really is and what's box score contribution?

[SideshowBob]

It's prior-informed RAPM that uses SPM as the prior for the offensive estimation. It's capturing +/-, but using SPM as a prior so there's a clear box-score bias. The bias on the defensive end is minimal; the prior is largely based off previous defensive RAPM, but they've somehow incorporated the box-score into that as well (though not to the extent that you'll see in SPM/ASPM).

However, we do have regular prior-informed RAPM for this season as well.

https://docs.google.com/spreadsheet/ccc?key=0AqnFn6ok911sdFVmQ3MxN29teWJkRXByQ0theUZnM2c#gid=0

I believe you commented on this a few weeks back. FWIW, v-zero has refined it a bit since then.

[Doctor MJ]

It's prior-informed RAPM that uses SPM as the prior for the offensive estimation. It's capturing +/-, but using SPM as a prior so there's a clear box-score bias.

However, we do have regular prior-informed RAPM for this season as well.

https://docs.google.com/spreadsheet/ccc?key=0AqnFn6ok911sdFVmQ3MxN29teWJkRXByQ0theUZnM2c#gid=0

I believe you commented on this a few weeks back. FWIW, v-zero has refined it a bit since then.

Okay, I'll have to check it out more. I do remember it, but at the time I didn't have enough confidence in it.

[SideshowBob]

And I agree with your views on the issue Doc. I guess most of the APBRmetrics guys are just trying to look for something different than we are. They're basically looking to make the best statistical predictions, over using statistics to perform analysis (not to say they aren't doing this, just that one is clearly prioritized over the other). For that reason, I can see where they're coming from, but it hurts the overall analytic community that lack the knowledge base or resources to come up with the kind of tools we need and have to rely on guys who are getting hired into silence to provide them.

[SideshowBob]

It's prior-informed RAPM that uses SPM as the prior for the offensive estimation. It's capturing +/-, but using SPM as a prior so there's a clear box-score bias.

However, we do have regular prior-informed RAPM for this season as well.

https://docs.google.com/spreadsheet/ccc?key=0AqnFn6ok911sdFVmQ3MxN29teWJkRXByQ0theUZnM2c#gid=0

I believe you commented on this a few weeks back. FWIW, v-zero has refined it a bit since then.

Okay, I'll have to check it out more. I do remember it, but at the time I didn't have enough confidence in it.

I've felt comfortable using it within the season, but I haven't been too sure that it should be compared to any of Engelmann's single-year studies because I can't imagine that somewhere there's a difference in the math (not to say that one is more correct than the other).

[NO-KG-AI]

Funny thing is, scoring in the pick and roll is a great thing. Teams have trouble taking the P&R away, that's why it's been a huge staple in basketball for so long. It's not like teams don't try to disrupt his P&R every single night.

If you take away his ability to get to the rim and finish, draw fouls, or hit the 3(and his ability to create for others), then he's got nothing!! Flawed!!

[SideshowBob]

Here's the page. He's got non-prior informed up as well.

[rrravenred]

Funny thing is, scoring in the pick and roll is a great thing. Teams have trouble taking the P&R away, that's why it's been a huge staple in basketball for so long. It's not like teams don't try to disrupt his P&R every single night.

If you take away his ability to get to the rim and finish, draw fouls, or hit the 3(and his ability to create for others), then he's got nothing!! Flawed!!

What've the Romans ever done for us?

#MontyPythonQuotesAreNeverGratuitous