Is James Harden Truly an Atrocious Defender?

There are currently hundreds of GIFs, Vines, and YouTube videos of James Harden's bad defense. And with the Kevin Love to Cleveland move, and the San Antonio Spurs stifling the Miami Heat in these past NBA Finals, defense is certainly all the rage right now.

However, it’s still an area of basketball where the "eye test" can directly conflict with a lot of the available advanced statistics.

There are several examples of this (Love is a good one as well), but let’s dive straight into Harden since he’s the topic of this article. Harden had a defensive rating, defined as points allowed per 100 possessions, of 107 this past season. That number isn’t great, but it seems to come with a lot of noise. It’s very teammate-dependent, and playing with better defensive players will obviously give a bad defender a better rating.

Some people will immediately write off defensive rating, especially when they see that some of the better guard defenders - like Avery Bradley, for example - posted a worse rating (109) than Harden. Even Harden’s defensively superior teammate, Patrick Beverley, had a worse rating of 108.

So let’s look at ESPN’s Real Plus-Minus, as it specifically attempts to eliminate the teammate issue for individual stats. The results seem to be more in line with the eye test – the best shooting guards per DRPM were Tony Allen and Danny Green. Harden ranked 74th out of the 86 rated shooting guards in DRPM, and it says that the Rockets allowed 2.84 more points per 100 possessions when Harden was on the court.

But here’s where things get really interesting. The great site nbawowy.com lets us set a specific player “on” and “off” the court, and filters the team data accordingly. This first table is how opponents performed against the Rockets with Harden on and off the court.

PPP = point per possession
TS% = true shooting percentage
OPPP = opponent's point per possession
OTS% = opponent's true shooting percentage

And here’s how the Rockets performed on offense with Harden on and off the court.

As you can see, Harden’s effect on the offensive end far outweighs the negative that he brings defensively. And it’s not really close at all.

(Interestingly, opponents actually shot worse when Harden was on the floor.)

Perhaps the reason for this is that Harden is playing with great defenders who provide a high defensive floor regardless of who else is on the court. While this may be true, in the idea is also the potential that perhaps defensive scheme and coaching is more important than individual players.

The caveat in the above statement would be with rim protectors – there's a growing sentiment that rim protection is far and away the most important defensive variable. And this doesn’t conflict with our Harden findings; perhaps wing defenders just aren’t that important to overall team defense. With a great rim protector like Dwight Howard, it doesn’t seem to matter that much how bad (or good, like we see with Patrick Beverley’s data) the wing defender is.

And even with that caveat, there's still a lot to be learned about team defense. Let's touch on the coaching and scheme topic with an example. Charlotte Hornets head coach Steve Clifford crafted a top-five defense last year with Josh McRoberts and Al Jefferson getting most of the frontcourt minutes.

This probably deserves an article unto itself, but what we can take away from it now is that we still aren’t perfect at predicting team defense. When you have almost the same roster in 2014 as you did in 2013, except for a non rim-protecting big in Jefferson, and you go from the league’s worst defense to the fifth-best one, it has to be coaching and scheme.

This means that perhaps we’re overstating individual defense in the league. The videos of Harden getting blown by are certainly entertaining, but the data just doesn’t show Harden crippling his team as much as many analysts may claim. He may be hurting the team defense, but it seems to be minuscule in comparison to how he helps the team offense.

So, is Harden a good defender? Of course not. I’m just not convinced it matters very much.