Time to take on my boy Rohan over at @ The Hive. He did an exceptional post the other day that started out with him trying to prove that CP3 was a good defender. Much to his chagrin, he came to the realization his stats proved Paul is an inferior defender. The article was an excellent evaluation of Paul using two stats - Opponent eFG%(FG% adjusted by 3-point shots) and Defensive Rating - which is used to determine the Adjusted +/- stat.(a complex stat that evaluates how the team performs as a whole defensively when the player is on and off the floor). It's well worth a read.
Now, I'm going to slam it.
I'll start with Opponent eFG%, it is true that Paul lets opposing PGs shoot at a higher percentage from the floor. The average eFG% of point guards is 49%. Paul lets his shoot 52%. There are a lot of reasons for the improved shooting percentage, almost all revolving around Paul's size - he roams for steals, he can't challenge shots that effectively, and he has trouble fighting through high screens, allowing open jumpers or drives to the hoop based on whether he trails or goes under.
What does Paul do to counteract this problem? He plays disruptor, generating steals, poking away the ball to teammates, and generally trying to wreak havoc. Is this an effective way of compensating? Many pundits would say no, claiming solid on-the-ball defense is more important, but here's some numbers:
- An average PG takes 15.8 shots in 48 minutes. With an eFG% of 49%, they produce 15.5 points per 48 minutes.
- A PG opposing Paul has an eFG% of 52%, allowing them to produce 16.4 points per 48 minutes.
So Paul's defense gives up an extra .9 points per game compared to an average PG. Here's what his compensation does:
- An average PG steals the ball 2.0 times every 48 minutes. A posession is worth 1.04 points on average in the NBA, so an average PG takes away 2.08 points from the opposing team.
- Paul steals the ball 3.4 times every 48 minutes. A posession is worth 1.04 points on average in the NBA, so Paul's steals essentially defend 3.54 points per game.
So Paul's defense takes an additional 1.44 points per game away from the opposing team compared to an average Point Guard.
The net, is +.5 points per game in Paul's favor, and that doesn't include the fact that a steal always results in a posession change - which adds points to the stealer's team, while a forced miss only results in a posession change 70% of the time.
The other part of the argument revolves around Defensive Ratings. Defensive rating is an excellent stat at telling you how a team performs defensively when a player is on and off the floor. The problem is what it actually compares is not how good a player is defensively compared to other players in the league, but with players on his own team. The site where this stat is readily available, basketballvalue.com, makes that disclaimer on its homepage.
Normally, that's not really an issue. Starters should be better at defense than their backups, but all season long, that's not been the case with the Hornets. Back in January I created a post about the Hornets reserves this season, evaluating their offense and defense as a whole unit. The verdict was that our reserves couldn't score at all, but were producing incredible numbers on the defensive end - generating a defensive efficiency of 88.0.(The Celtics lead the NBA with a defensive efficiency of 96.1 right now) So the Hornets backups are great defenders - which is pretty obvious when you watch the team. Those stretches during the last part of the 1st and 3rd and early 2nd and 4th are ugly. No one scores.
So the result? The numbers say when Paul leaves the floor the team's defense gets better, and it also says the same for every single Hornets starter except for David West, who just barely improves the defense - probably because the standard rotation keeps him on the floor with the second unit for longer than anyone else.
According to the Defensive Rating numbers, our best defenders this season are, in order, Marcus Vinicius, Melvin Ely, Rasual Butler, Bobby Jackson, Ryan Bowen, David West, and Jannero Pargo. Everyone else on the team makes the defense worse. In other words, Paul, Stojakovic, Chandler, and Peterson play starters minutes and contribute to a top 5 league defense, but they all suck. Red flag, anyone?
The primary conclusion you should draw from the Defensive Ratings of the Hornets is that the Hornets starters defend the opposing team's starters less well than the Hornets 2nd unit defends the opponent's 2nd unit.
Now, am I claiming Chris Paul is a great defender? No. I would suggest he's about average overall. His size limits him, no doubt, but his speed and hands make up for the difference.
Normally, I'd move on to a Game Preview for tonight, but the above post took so much time to research and put together, I'm not going to have time for one of my full treatments.
The Clippers are almost as dangerous as the Kings, and in case you don't remember, last year we spoiled their playoffs hopes by beating them twice in the last couple weeks of the season. Brand will probably have vengeance on the mind.
We'll need to come out hard to win this one. Hopefully we do, and we can get some rest for our starters in the fourth quarter.
One can hope, eh?


16 awesome comments post your own
mW
04/15/08 07:31 AM
Interesting stats, Ryan. It seems that CP does indeed at least make up the gap with his speed and agility, in what he loses in strength and height. But he's only 23. Imagine what the crafty bastard will be like in a few years...
www.hornetshype.com #1
atthehive
04/15/08 07:48 AM
Great stuff, Ryan. My dream is that on this same day next year, we can both put up posts describing how awesome his defense is.
atthehive.wordpress.com #2
Ron Hitley
04/15/08 07:50 AM
Great stuff indeed. It's pretty tough to measure something like defense with statistics, but I like the conclusion here. That is, there's no black and white, but shades of gray.
www.hornets247.com #3
Robin
04/15/08 08:30 AM
I don't watch the team much, but I had the impression that rookie Julian Wright is a pretty good defensive player. Does he contribute to the second unit defense at all, or am I wrong?
#4
atthehive
04/15/08 08:42 AM
Robin, According to the basketballvalue data, he is "bad." According to opponent counterpart production, he holds opposing small forwards to a 13.9 PER... which is outstanding. Amazing, you could say. Just from watching him this year, he has very good footwork, great sense of positioning on the court, and I'm positive he will become the team's BEST one on one defender in one to two years. That's how good he is.
atthehive.wordpress.com #5
Ryan Schwan
04/15/08 08:51 AM
I agree with Rohan. Julian still sometimes makes a bad rotation or rookie mistake on defense, but as a one-on-one defender taking his man, he's easily the best our team has. He makes good use of that freakish giraffe calf length.
www.hornets247.com #6
Mikey
04/15/08 09:06 AM
Not to mention those mad giraffe calf hops!
#7
Jason
04/15/08 10:18 AM
One question about the defensive PER stat. Shouldn't you be taking into account the quality of opponent the player typically guards? If (for example) Julian Wright holds his opponents to a PER of 13.9, but the typical player he guards has an average PER of 12, he would actually be a bad defender, wouldn't he? Shouldn't you actually be figuring out the PER he holds opponents to in comparison to their usual PER, rather than in comparison to a league-wide average?
#8
Stefan
04/15/08 10:37 AM
If we lose to the Clips tonight, I will lose my mind!
#9
TheSecretWeapon
04/15/08 11:05 AM
Not to get too technical, but the counterpart data doesn't actually tell us that Paul permits opposing PGs to shoot a higher efg than the league average. The more correct way of stating that is this -- Players designated to be PGs by 82games shoot a higher eFG when Paul is in the game and designated to be matched against the player identified as the opposing PG. I know that sounds remarkably eggheaded, but well...I'm kind of a stat egghead. My main point is that there are significant issues with the counterpart data. It doesn't mean what we'd like it to mean. Essentially, counterpart data is generated by making a list of each team's roster from "most PG" to "least PG". The system automatically assigns positions based on predetermined designation. It assumes that the player identified as the PG for one team is matched up against the player identified as the PG for the other team. And so on through the positions. There's no accounting for help defense, which is AT LEAST half of a defender's responsibility -- more for a big. There's no accounting for switches or zones. There's no accounting for crossmatching -- the PG being assigned to defend the SG, etc. A single error in the matchup automator means there's a 40% error for that lineup. In other words -- there are significant problems with this data. Outstanding analysis on the rest of it, by the way. Out of curiosity, how did you go about calculating defensive rating for the bench?
kevinbroom.wordpress.com/ #10
Emir lidan
04/15/08 11:15 AM
wait peja's bad at defense didnt we just argue about this a couple weeks ago after that writer said peja was like a traffic cone or whatever pretty sure we all agreed peja was a good defender
#11
atthehive
04/15/08 11:22 AM
@SecretWeapon, You're not being eggheaded at all, that's exactly right. You hit every flaw in the data on the head. The problem is that the counterpart data, and the adjusted +/- are THE best defensive statistics that pretty much anybody anywhere has. Only team insiders track specific opponent movements on the court, and even in that case, I believe only about 30 to 40% of teams do it extensively. The way defense is currently measured is highly flawed. But the general thinking is that though this error exists, to an extent, it exists uniformly. Teams that are more eccentric in their defensive strategies will be more susceptible to higher error (and vice versa). So we need to look at the data, and look at what makes sense. Of course, that leads to more problems- what if we're leaving out data we "think" don't make sense, but actually do? As far as help defense, adjusted +/- is really the only thing that helps, but that stat has a long way to go before it can be fully believed. For example, adjusted +/- tells us that CP's bad defense cancels out his good offense. If that's the case, how are the Hornets 2nd in the West? If their best player is making a net "zero" contribution, how is that statistically possible?
atthehive.wordpress.com #12
Ron Hitley
04/15/08 11:28 AM
@ Emir: Ryan didn't say Peja's bad at defense here. Note the "red flag, anyone?" @ TheSecretWeapon: I left a similar comment over at At The Hive. Statistical analysis is limited and can only tell us so much. I think a good illustration of your point is when the Hornets played the Warriors this season, and Paul usually guarded Monta Ellis rather than designated G-State point guard Baron Davis. In that case, would Baron's FG% against the Hornets be considered his FG% against Paul, even though they didn't guard each other very much?
www.hornets247.com #13
sparks
04/16/08 06:21 AM
i think that stat is wrong. i read an article on hoopsvibe that kobe bryant and chris paul make the person they are guarding shoot 3 and 4 percent less than the average guard - and then kevin garnett makes the opposing power foward shoot 13 percent less than average!!! we know kevin is one of the best defenders of all time, and kobe does his thing for a few years now. so i think that stat doesn't really tell the whole truth about chris paul
#14
GHJansom
04/17/08 01:22 AM
Geetings Wanting to spend some dollars in http://diamonds.yuchaszsports.com/sitemap/>jewelry travel boxes will you guys, know any honest online jewelry stores? Thank You In Advance
#15
TheSecretWeapon
04/21/08 06:56 AM
Team "insiders" aren't the only ones tracking. I've tracked Wizards games off and on since 2005 (not much tracking this season, though -- I decided I wouldn't do it unless I had a client, and I didn't have a client). Hand tracking data is definitely a step in the right direction, but it takes time. That time comes down with experience in tracking, but more in-depth (and useful) stuff still takes both time and knowledge. I think the team on/off stuff is better than the counterpart data. To me, the counterpart stuff can be interesting at times, but I don't think it adds much analytic insight because of its inherent flaws.
kevinbroom.wordpress.com/ #16