State of Iguodala and Thaddeus Young's Defense

[76ciology]

http://www.basketballprospectus.com/art ... icleid=918

It's not uncommon to read a lament about how the NBA's All-Defensive team is dominated by star players with sparkling scoring averages. Whether or not that actually happens is a matter of debate. It seems like more often, the coaches make poor choices based on gaudy shot-blocking or steals totals, but with names like LeBron James and Kobe Bryant cropping up annually on the All-Defensive squads, there's no question that some of the game's most explosive scorers have been honored for their work on the other end of the floor.

I have no problem with this. In baseball, there is an oft-discussed problem with great hitters winning undeserved Gold Glove Awards. (*cough* Derek *cough* Jeter *cough, cough*) Some of this backlash may have found its way into many people's response to NBA defensive awards. It shouldn't.

The best all-around offensive players in the NBA will almost invariably be the top defensive players in the league because they're tapping into the same skill set. Length, quickness, strength, leaping ability and desire are just some of the traits that differentiate the best players no matter which end of the floor we're talking about. Defensive specialists such as Bruce Bowen or Jared Jeffries can be terrific athletes but unskilled offensive players. They are able to carve out well-deserved reputations for defensive prowess by finding a niche and a being willing to accept that role. But even the best defensive specialists don't have the defensive ceiling of the NBA's transcendent athletes--James, Bryant, Dwyane Wade, Dwight Howard and, as Kevin Pelton pointed out in his latest piece on the Thunder, Kevin Durant.

I bring this up because today I'm going to list the players designated by NBAPET as this season's top defenders. The top of the list is star laden, reading very much like a rundown of the most valuable overall players in the league. Again, I don't have a problem with this, though there are some peccadilloes with some players on the list which I'll point out.

When you present a list like this, you always have to run off a string of caveats and explainers, as if I were Thomas Haden Church doing a voiceover for a shady pharmaceutical company. (Aren't they all?) Anyway, here goes:

1. The key defensive metric in NBAPET's rating of individual defensive prowess is dMULT, which is a non-creative way of abbreviating Defensive Multiplier. The bad name evolved from the role it played in NBAPET's system of rating overall player value. NBAPET wants to measure the differences in efficiency and production from what a player gets and what he allows. In one early incarnation of NBAPET, this resulted in a multiplier that mashed together several different approaches of measuring individual defense and was then used to "adjust" a player's offensive production. I dropped that methodology about five years ago, but the name dMULT remains.

What dMULT now attempts is to measure is the points created per possession used allowed by a player's box score counterparts versus their usual production. So a dMULT of .869, which is what Thabo Sefolosha currently sports, means that his counterparts have been 13.1 percent less productive when going against Thabo. A dMULT of 1.105 would mean that opponents have been 10.5 percent more productive. And so on. (Current dMULT scores are listed on our player pages.)

2. I make no claim to this being any sort of end-all/be-all of individual defensive metrics. The system doesn't work particularly well for measuring the defensive ability of reserves, and the numbers are heavily regressed for playing time to smooth this out. But that also means that a player with little playing time is going to rate close to average. Also, I have long been dissatisfied with the box score counterpart approach and, as such, I've been (very) gradually working on converting NBAPET to a play-by-play based system which tracks estimated opponent data on a possession-by-possession basis. (For me, it's a big project because I lack the programming skills needed to simply write a play-by-play parser. I have to use Excel for everything.)

3. In NBAPET's system of skill ratings, all players' dMULT scores are compared to their positional averages and then all players are ranked according to these computations. From there, a percentile ranking is assigned. Finally, the percentile ranks are fitted into a Bell Curve to give each player an integral rating between +5 and -5, with fewer players landing ratings on either end of the curve and more players clustered around the middle.

4. Even when I review skill ratings for defensive ability, I always look at them in conjunction with ATH (a rating for athleticism), BPS (position-adjusted blocks plus steals) and on court/off court stats. If all of these measures confirm what dMULT is telling me, then I'm much more prone to declare a good or lousy defender. If they don't support dMULT, then I start fishing for an explanation.

With that out of the way, let's move on to the list of the 17 players currently rated as +5 defenders by NBAPET, in order of dMULT.

1. Dwight Howard, ORL (C, .709): There is a near consensus that Howard has become the NBA's most dominant defensive force. Physically, he has it all and he really exhibits the kind of mobility it takes to be a great defending big man in today's NBA.

2. Kobe Bryant, LAL (SG, .732): I've never been able to think of an approach to adequately address this issue, but I suspect that the outstanding defensive numbers posted by some of the game's top scorers can be attributed in part to their ability on offense. Whether it's energy, foul trouble or what, the task of defending these guys bleeds over to the offensive end. In any event, you won't hear many arguments against Bryant's defensive skills.

3. Kevin Durant, OKL (SF, .738): He has every tool, he just needed the want-to and the support offered by better surrounding talent.

4. LeBron James, CLE (SF, .771): Once LeBron decided to make defense a priority, it was lights out for Cavaliers opponents.

5. Dwyane Wade, MIA (SG, .796): Wade is a player I'd point to as one that holds down opponents production in part by making them really work on defense. He also fills up the defensive box score with steals and blocks.

6. Luol Deng, CHI (SF, .814): Deng is underrated in many ways, but perhaps no more so than on defense. Having seen him play many times this season, I can attest that my eyes confirm what the numbers suggest. He gives Carmelo Anthony fits, just to cite one example.

7. Andrei Kirilenko, UTA (SF, .816): Kirilenko has always sported elite BPS and ATH ratings, but his dMULT has sometimes lagged. My theory is that he's had too much of a tendency to roam looking for defensive swag. This season, that hasn't been a problem.

8. Dirk Nowitzki, DAL (PF, .817): Nowitzki had a terrible defensive reputation when he was a young player in the league and I think that still colors many people's perceptions of his effort on that end of the floor. He's been consistently solid in dMULT the last few seasons, though never to quite this degree.

9. Kevin Garnett, BOS (PF, .842): Even with diminished athletic indicators and reduced playing time due to injuries, KG still rates as an elite defender.

10. Al Horford, ATL (C, .849): A dedicated and cagey post defender, Horford will garner support for second-team All-Defensive this season.

11. Ron Artest, LAL (SF, .858): Artest was a little annoyed at making "only" second team All-Defensive last season and he could be headed towards a repeat performance. He's been really good, as usual, but there have been other forwards having better defensive seasons. Incidentally, Trevor Ariza, who Artest replaced on the Lakers, has a +3 defensive skill rating for the Rockets.

12. Thabo Sefolosha, OKL (SF, .859): Sefolosha's complementary game makes him an ideal fit with Oklahoma City's first unit and he teams with Kevin Durant to give the Thunder a pair of top wing defenders.

13. Hedo Turkoglu, TOR (SF, .869): This one seems suspect, given Toronto's overall showing on defense. The only other Raptor with a positive defensive rating is Marco Belinelli. At the same time, Turkoglu had decent dMULT figures the last couple of years in Orlando. In the end the Raptors have been better defensively when Turkoglu has been on the bench, so chances are that he's getting at least some credit for somebody's else work.

14. Joe Johnson, ATL (SG, .870): Johnson has not always rated as a good defensive guard, but he experienced an uptick last season that has continued in 2009-10. The Hawks are in the middle of the pack defensively, but Johnson, Horford and Josh Smith give them a solid defensive core.

15. Lamar Odom, LAL (PF, .870): Odom's combination length and mobility is truly rare and his dMULT is backed up by the fact that the Lakers' Defensive Rating improves by 5.5 points when he's in the game.

16. Caron Butler, WAS (SF, .870): Butler's year-to-year consistency in dMULT has been nil--you never know what you're going to get. If his rating is real, however, and he ends up going to Boston as rumored, he'll fit right into Doc Rivers' defensive schemes.

17. Danilo Gallinari, NYK (PF, .873): When I've watched the Knicks, I haven't really taken any special care to watch Gallinari on defense. Next time, you can bet I will. I don't particularly buy this rating and suspect that there are some cross-matching issues involving Jared Jeffries, who does not fare well in NBAPET's defensive ratings.

Based on this research, he's no better than guys like Odom, Hedo Turkoglu, Gallinari and Joe Johnson.

Based on On/Off defensive rating:
(Weighted) On and Off Court +/- per 200 Possessions. Weighted +/- in an approximately average length NBA game when player of interest is on the court for their team or teams. Weighting is based on minutes played, not possessions, so as not to overweight play for higher tempo(possessions per game) teams.

2009-2010
Thad Young: -1.25
Andre Iguodala: -.42

This just goes to show that Thad Young made three times more defensive impact when he's on the court compared to Iguodala.

2008-2009
Thad Young: -7.42
Andre Iguodala: -7.71

This just goes to show that both Thad Young and Iggy had a very good defensive impact when they are both on court. -7.++ is pretty impressive for on/off def ratings. That puts them in the top 4 and 5 spots, just below LeBron, Joel Przybilla and Marcus Camby and above Ben Wallace and KG.

Overall: Is Iggy overrated and while Thad is underrated, in terms of defense?

*If you guys know what other defense related statistics to compare both players. Please share.

[Point-God]

Coaching has a great deal to do with the numbers and the numbers can not account for that. A coach will normally place in defenders the most advantageous position. For instance, it is advantageous for the coach to assign D. Wade to defend the opposition's worst offensive player between the two wings in order to conserve his own energy for the offensive end of the floor. It is advantageous for the coach to assign Iguodala to defend the best of the opposition's two wings because part of Iguodala's impact is holding players at or below their average. Lastly, with a reserve tweener like Thad, the coach is looking to assign him defensively to whatever player will give Thad the least amount of problems, and if the matchups aren't good, Thad may not play much that game... That's why all of those numbers above are garbage. The element of coaching is very important...

[corwin]

I have no idea what any of those stats really mean. I love Thad's potential but I was pretty disappointed watching his ability to stay in front of his man last year. I'm waiting to see how he performs with a coach stressing the fundamentals of the game.

[freshie2]

Interesting, but I have no idea what it means at this point. If they are in the top 5 (in a good way) in the league in this area, then I'd say there is something to be said for having them on the court together...not sure what else you can say on this.

[ChuckS]

When you present a list like this, you always have to run off a string of caveats and explainers...

I make no claim to this being any sort of end-all/be-all of individual defensive metrics. The system doesn't work particularly well for measuring the defensive ability of reserves, and the numbers are heavily regressed for playing time to smooth this out...

In NBAPET's system of skill ratings, all players' dMULT scores are compared to their positional averages and then all players are ranked according to these computations. From there, a percentile ranking is assigned. Finally, the percentile ranks are fitted into a Bell Curve to give each player an integral rating between +5 and -5, with fewer players landing ratings on either end of the curve and more players clustered around the middle...

Even when I review skill ratings for defensive ability, I always look at them in conjunction with ATH (a rating for athleticism), BPS (position-adjusted blocks plus steals) and on court/off court stats...

Some years ago statisticians became famous for systems measuring individual excellence in baseball. I found little fault with most of their findings, however, because generally the same players are on the field throughout the game. There were some problems with relief pitchers, but generally you can expect an ERA, for example, to be fairly accurate, and determinative between those who are facing the same opponents throughout the long season.

Trying these techniques in basketball, however, created what I think is the worst thing that ever happened to fan understanding. I'm not worried about the professionals. They've played, have their own scouts, and will use the numbers as an indicator, or at best a starting point or factor in overall analysis.

Unlike baseball, and to a somewhat lesser degree football, a basketball game has an infinite number of variables, with up to twenty four players on offense and defense for 48 minutes. Some are playing predominantly against starters, some vs. reserves, and some, for some minutes, against both. Compound that with the myriad defenses which remove direct responsibility, like zones, trapping, switching, doubling, etc. For these and more reasons I do not believe any mathematician will ever devise anything but, at best, risky guidelines, regarding defense.

Read their own caveats, and look at what convoluted weightings they must add, based on their own lay (usually) beliefs of what they think important or necessary, to even develop a system. It is my, probably singular (at least on this board), opinion that the 48 minute protraction is the biggest perversion of real numbers. At least I thought it was. This guy is more vague: "The system doesn't work particularly well for measuring the defensive ability of reserves, and the numbers are heavily regressed for playing time to smooth this out...". What the hell did he do?

This just goes to show that Thad Young made three times more defensive impact when he's on the court compared to Iguodala.

I rest my case.

[76ciology]

Chuck,

I know those stats aren't perfect. For instance, you can't find out who's has a more defensive impact between two teams by just looking at their on/off defensive ratings, for different teams plays different styles/philosophies in the defensive end. But if you're comparing two guys' on/off defensive ratings between two guys on the same team, given that they have played major minutes, I think you can see who has more impact at the defensive end.

None the less, I think that this studies are good indicators if such player is a good defensive player or not. Because guys who are said to be good defensive players, specially the more obvious ones, usually have "good stats" when it comes to this statistics.

[PhillyRocks1]

The Bobcats had the best defense in the league. Wonder why none are listed on this?

[Sixersftw]

Chuck,
But if you're comparing two guys' on/off defensive ratings between two guys on the same team, given that they have played major minutes, I think you can see who has more impact at the defensive end.

I dunno if these two are comparable with advanced statistics defensively (which I feel are majorly lacking). Andre almost always guards the best perimeter player on the court where Thad guards the worst. Though they both played major minutes Andre played over 1000 more minutes than Thad. I question how reliable a stat based on playing time is when comparing a player that was on the floor 80% of the time versus a player who was on the floor for 54% of the time. Like you acknowledged, there are just too many variables in play when comparing players w/ this stat, or at least ONLY this stat.

That first article is a bit convoluted but I would question Gallo as the 70th ranked player defensively let alone the 17th

[PhillyRocks1]

I love new stats that are made up for some stupid reason.

[ChuckS]

None the less, I think that this studies are good indicators if such player is a good defensive player or not. Because guys who are said to be good defensive players, specially the more obvious ones, usually have "good stats" when it comes to this statistics.

It is my belief that "cream rises to the top", and no system can make some players look bad.

Hollinger and Beech, who I think are probably the best of these new age statisticians, although I disagree with some of their algorithms, and of course 48 minute projections, prove your point, and mine I think.

I generally have less disagreements with the top ten on their lists. But occasionally even there, and definitely throughout the rest of their compilations, I find numerous unbelievable, and sometimes even laughable, ratings.

The biggest problem, as I see it, is any method of trying to overcome the disparity of minutes, and isolating individual performances with infinite variables.

I am not implying that the systems of the better statisticians are useless. I just, personally, have more faith in numbers compiled from box scores, or better yet using the box score for any game to supplement my personal observations, or to suggest that I may have missed something. Of course these numbers have their limitations. I just do not believe they can be as far awry as protracted numbers with questionable weightings.

In any event, any system that suggests that Thad is even remotely comparable to Andre defensively is quite simply wrong. If it in any way suggests that he is better, then its finding is ludicrous.

[bballin76]

waiting for Derek.

[Sixerscan]

I bet that there is some serious fluctuation year to year with this statistic. All encompassing defensive statistics in basketball are a long way away.

Regardless, anyone with two eyes that watched Hedo or Gallinari play will tell you they aren't as good defensively as Iguodala or Young. This doesn't pass the laugh test.

[F-Stop]

it's real simple. The guys who are high scorers (Wade) often get there starting counterparts in foul trouble early thereby reducing a 32min 15pt player into a 27min 12pt player. The back-ups are then given more minutes during that particular game and you end up the 35-40pt game while the opposition has to settle for what the bench can muster. Iggy being the best defender and best player on the team has to take defensive possessions off so as too stay in the game or his 40min nite is reduced to 32min thereby giving the 6ers almost no chance to win.

[SJSF]

ANother reason why i say Squiggy needs to go. I never saw his defensive prowness as anything great. If anything i thought he was average at best on the perimeter. Thad Young i still believe is only average.

[Kobblehead]

at topic.

[SJSF]

it's real simple. The guys who are high scorers (Wade) often get there starting counterparts in foul trouble early thereby reducing a 32min 15pt player into a 27min 12pt player. The back-ups are then given more minutes during that particular game and you end up the 35-40pt game while the opposition has to settle for what the bench can muster. Iggy being the best defender and best player on the team has to take defensive possessions off so as too stay in the game or his 40min nite is reduced to 32min thereby giving the 6ers almost no chance to win.

U lost me at best player. Also Squiggy never guards the teams other best player.

[HotSauce77]

Iguodala is an excellent perimeter defender and a very good defensive player overall. Young is below average at both PF and SF defensively. He has more potential defensively at SF and should at least work himself into an average defender. If tasked with less offensive and ball handling responsibility, Iguodala can be an all-defensive team performer. And yes, Iguodala is the guy we use to guard Wade, to guard Pierce, to guard LeBron, to guard Kobe, etc.

[ZzAzZ]

No offense to the OP but that was the worst list of "good defenders". Turkoglu was on there for goodness' sake.

[PhillyFan]

OMGWTFBBQ, lets trade Iguodala for Luol Deng since Deng is rated 6th in the league for his defense!

[ZzAzZ]

Also Squiggy never guards the teams other best player.

Yeah, that's dead wrong (apologies if you are being sarcastic).