New advanced defensive metrics: James Harden was a poor defender in '13-'14

I remember there were some deniers of Harden's horrible defense last season, citing some random stats to argue that those of us who say Harden's defense was poor were just haters. I am glad that Harden faced the truth about his terrible defense unlike some of his apologists here.

 

I got bashed hard when I said Parsons and Harden both regressed. Glad the stats are up now for the stats only can see.

 

Exactly, the Blazers went straight at him with Matthews post ups last year.

 

Harden for MIDPOY (most improved defensive player of the year).

 

Yeah, when a stat spits out a bunch of bench players at the top of the list, you have to question it.

 

Miserable fail at Deadspin twitter sharing this article and spinning it making it seem like he is still bad on defense this season. I mean my god I know its the internet I know people are going to troll, I do it all the time, but this is getting really fkn pathetic. I cannot wait until James Harden hoists that NBA finals trophy and hits the Shmurda dance and shut all these fools up.

 

If the metric told you Harden was a great defensive player last season, would you have believed it? So if it disagrees with you, you say it's got to be wrong. If it agrees with you, you say it tells you nothing you don't know. Basically, you trust your eye test as the standard to judge stats, not the other way around.

I've always wanted to see some kind of meaningful metrics to measure individual defense. I don't know if this is it. But I welcome the attempts to find it.

 

Zach Lowe made a pretty good point about his conditioning. Last year Harden would take plays off every 3 or 4 minutes, which would lead to well...yeah. This year he's taking an even greater load of the offense, and in spite of that he needs to be running the offense and playing d for 15 minutes straight before he takes one off (and at that point he deserves it)

 

This is about defense. We all know that offense makes stars. I wouldn't be surprised that many bench players were better defenders than stars.

 

This confirms what I have seen by watching Nick Calathes. He's headed to restricted free agency and could possibly be an underrated asset acquirable on a value contact.

 

CH, please keep in mind that a fair number of "Harden apologists" may have been motivated to defend him due to your relentless Harden-bashing last summer. I'm glad you have found other things to talk about this season.

 

Wait, I'm confused. Are these new defensive stats or year old defensive stats?

See, that's the problem with analytics, they can't help us see present trends, rather it takes them 6-12 months of data to find trends.

 

I'm sure Rockets are using these metrics to track our defense. You always see Finch and sometimes McHale poring over these reams of paper. They are stats they're looking at. Sometimes you see Finch talking with Harden while looking at those stats.

Also every team has those SportsVU cameras installed and have access to the data through the SportsVU company.

Does anyone know if teams can access other teams' stats automatically? Or just confined to their own?

 

Dwight so overrated tho. Rockets better without Dwight tho.

 

The data is there right now - they can tell you who the best defender was in the last 10 minutes tonight, over the last week, the last month, etc. The reason you don't see that data about present day is teams want to keep a competitive advantage. They are just illustrating a point and old data is fine to prove the point.

 

The one problem I had with that Goldsberry piece(and I told him on twitter afterwards), is the example of Chris Paul guarding Steph seems to gloss over the fact Deandre Jordan was waiting in the paint preventing Curry from speeding past Paul on the closeout and thus allowing Paul to recover in time to contest the jumper. I thought that was a great example of how a shot blocker in the paint forces players to shoot more midrange shots - but not Paul's defensive brilliance. Paul is crafty, gets a lot of steals, and has a lot of respect from opponents but I struggle to see him as a lockdown defender. Maybe at one time, but the last couple of years he has not passed the eye test for me. Not saying he couldn't still be a good defender, but to me he takes off plays like Harden did last year.

 

I told the truth. Fine to defend Harden. Fine to argue that he could improve (as he did), but arguing that his D was just fine was delusional regardless of motivation.

Harden's D no longer stinks, so I no longer say his D stinks.

 

Okay. Well, we better make an adjustment based upon our advanced analytics of the 2015 draft based on the last 10 minutes of player data in the NCAA Tourny, too....right? The good teams are keeping the advanced data on March Madness a secret, right?

That's like saying you better buy or sell on the last 10 minutes of advanced analytics of stock data.

 

Wow - Barkley is that you?

...the way it works is like this:
- you get a big bunch of data
- you analyze the data looking for patterns
- you develop a statistical model based on the patterns you find
- you then apply your statistical model to new data as you get it
- and finally you periodically review the results to make sure things aren't deviating in a way that warrants tweaking or changing your statistical model.

All these NBA analytics guys have a statistical model for each important metric they define. They all arrive at those models based on their own research so every team's model is going to be at least a little different from each other if not completely different. No one is stupid enough to make any inferences based on such a small time of data as 10 minutes, my point was to answer your original question by saying that teams can certainly take recent data and add it into their data set to tell things for any length of time they define. Maybe they want to know who the best rim defender has been in the last 82 games starting with tonight's games? Maybe they want to see if a player's metrics are trending up or down. In either case, recent data feeds a larger data set and a defined statistical model to show value and trends.

 

half of those were probably jj redick...