It's looking at Harden's cumulative Player Impact +/- ... I'm not sure what that is exactly, but some kind of +/- stat adjusted somehow. Harden does lose us points defensively. This isn't new or a shock, or shouldn't be. I think more than anything it was trying to show how offensively he wasn't himself from a +/- impact, until Melo left. Then something else happened around game 17 that cratered Harden defensively. That was probably CP3 getting injured. He missed games 18-20. He was also suspended games 3 and 4, where there is a slight dip in Harden's defensive +/-. CP3 is important to Harden defensively... that's obvious, though.
Yeaaa, okay that makes sense. Thanks for explaining this for me, I didnt really get it at first. Sometimes I wonder if Harden is actually that bad of a defender than I see that he can actually move his feet laterally and have some good defensive possessions and then I get annoyed.
When CP is on the floor without Harden....that equated to the "bench unit." Stat backs up what we already know about that $hitty bench unit.
I've been looking at on/off too morning as well. I found some pretty weird stats. I was thinking about making a thread on what I found, but now I'm thinking that I shouldn't because my findings are different. I used stats.nba.com, so I honestly don't know if it's reliable compared to what that guy used. NBA says that CP3 lead line ups without Harden are +4.9 per 100. That guy has it at +2.6 which is a huge difference. Still, I'll put some quick findings (w/o the source numbers) in the spoiler below. Spoiler 1. James Harden and Chris Paul DO NOT work well together at all this season. Last season they were at least +10 per 100 when together. This season they are around -2.0 per 100. 2. Eric Gordon and Chris Paul DO work very well together. They are around +7 together, and work well with every starter except for James Harden. Although I've only checked CP3/EG line ups with the starters, and not with the bench players like Clark, House, and Green. But CP3/EG do work very well together. 3. James Ennis is bad across the board defensively and offensively, but when Harden and Ennis share the court then the team becomes bad. They are -7 per 100 together. Harden without Ennis is +5 per 100. Overall, James Ennis in the starting line up may be hurting the Rockets a lot more than we think. 4. Rolling off of #3, the starting 5 are -9.1 per 100 together. However, in the few minutes that Ennis isn't on the court the starting 5 minus Ennis are +5 per 100. 5. CP has been bad offensively without EG. He has a 99 off rtg without EG. However, Chris Paul line ups are typically the best defensive line ups for the Rockets 6. Harden led line-ups are typically the worse defensive line ups for Houston. However, Harden's offense is able to break-even with his defense because his line-ups are some of the better offensive line ups for the Rockets 7. Essentially, CP line ups bring the defense, while Harden line ups brings the offense. They don't bring the other though. The offense gets better when CP3 is off the court, and the defense gets better when Harden is off the court. Together, they are bad as revealed in #1 of this list. 8. PJ Tucker is very important, but because of his off court numbers. He leads the team in minutes played, but the Rockets almost break-even when he's on the court. However, in the limited time that Tucker is off the court both the Rockets defense and offense go to crap with a net rating of -7.5 per 100. 9. Capela has not been the defensive player we need him to be, but to be(aka fair I need to dig deeper on his on/offs. I need to dig deep on Gordon's on/offs as well. edit 10. I forgot this one. CP3 lead line ups are -16 per 100 when Harden and EG are not on the court with him. (aka bench) Harden led line-ups are -11 per 100 when CP and EG are not on the floor with him. (aka bench)
@oogie boogie that post was fantastic! #1 and #3 in particular. And I think #3 might translate more than we know in regards to the huge rebounding problems. And this team is going nowhere if the CP/JH duo is not augmenting each other. 1. James Harden and Chris Paul DO NOT work well together at all this season. Last season they were at least +10 per 100 when together. This season they are around -2.0 per 100. 3. James Ennis is bad across the board defensively and offensively, but when Harden and Ennis share the court then the team becomes bad. They are -7 per 100 together. Harden without Ennis is +5 per 100. Overall, James Ennis in the starting line up may be hurting the Rockets a lot more than we think.
Imagine James Harden running much more off the ball. I think it would be much better fit. James Harden uses deceleration very well.
I'm no Melo fan but you can't tell a story about Melo in the team stats due to so many collinear variables (Gordon's play, awful 3 point shooting, lack of Bzdelik, Paul's suspension/injury, Harden's missed time, awful bench play, etc... not to mention the small sample size with Melo that was way too early on the learning curve.) Better to just let that story fade away, I think.