clock menu more-arrow no yes mobile

Filed under:

I Know It When I See It

I am continually amazed at the voting for Defensive Player of the Year and the NBA All-Defensive team.  More than any other category, the voters seem (to me at least) to get this wrong, sometimes ludicrously so.  Interestingly, there are two different panels voting on these awards - media in the case of the DPOY, coaches for All Defense, so the problem goes beyond simple issues of bias (or rather, if bias is involved, it is fairly wide spread).

Part of the problem of course is that American sports culture is accustomed to having statistical metrics for measuring performance.  Who's the best pitcher in baseball?  Wins and losses, ERA, strikeouts, quality starts and myriad other new fangled stats help you figure it out.  Who is the MVP in the NBA?  Reasonable people can disagree, but the list starts with players who rank near the top in one or more statistical categories (points, assists and rebounds being the most important ones) and who play for a top team.  

The same is true, more or less, when discussing defensive awards.  Unfortunately, the statistics commonly used to identify candidates are at least flawed, and arguably meaningless.  

An interesting study by David Nelson and Damien Walker is available on currently.  To their credit, they broke down the stats that are currently used to distinguish top defenders, and then added in some additional stats.  Unfortunately, although they were cognizant of some of the shortcomings in this method, they were completely undeterred in trusting them to identify the defensive player of the year (they came up with Shawn Marion, a fine choice by the way).  

So what's wrong with the statistical approach?  Just that statistics and defense don't mix.  Usually I'm a big fan of stats, but they can certainly be misused.  (A recent study of an abstinence program showed that high school students were waiting significantly longer before having their first sexual experience - only problem was, it was only counting the data for the students who stayed in the program.)

Here are the stats that are commonly used to measure defenders, and the problems with them (imho).

Blocked shots - Blocked shots are what they are.  They are clearly a defensive stat, and they are clearly part of an effective defense.  Of course the category is dominated by centers and power forwards, who are long enough to guard the rim, and who generally have a defensive assignment that allows them to stay close to the basket.  Of the last 16 Defensive Player of the Year Awards, 14 of them were given to Marcus Camby, Ben Wallace, Dikembe Mutumbo, Alonzo Mourning, Hakeem Olajuwon and David Robinson.  Ron Artest and Gary Payton are the only non shot-blockers to have snuck onto the list since 1991.  But blocked shots are only one aspect of defense, and most blocked shots come from the weak side.  You can be a great shot blocker and a poor on-ball defender.  In fact, some shot blockers (not all) ignore their primary assignment and roam the paint looking for the rejection.  At any rate, if the NBA wanted to hand out an award to the top shot blocker on the season, they could do that, and it would almost eliminate the need for a DPOY Award as currently constituted.

Steals - Even more so than blocked shots, while ostensibly a defensive stat and a positive thing for your team's defense, steals are often times indicative of POOR defense.  Steals rarely occur in on the ball situations.  Most steals happen in passing lanes or on blind side double teams, and are the result of risky defensive practices.  It's a skill like anything else, and some people do it better than others.  And some players do it well, while also playing solid on ball defense.  But consider this:  Caron Butler was third and Gilbert Arenas seventh in steals this season in the NBA, yet the Wizards as a team were 28th out of 30 in points per 100 possessions.  If steals were a reliable indicator of terrific defense, then how could the Wizards be so poor defensively with too such stalwarts?  Answer: steals aren't a valid indicator.  Butler and Arenas take lots of chances on defense, which results in some steals, and lots of points for their opponent.

Defensive rebounds - I'm not really even convinced that voters use this category.  Nelson and Walker seem to think they do.  And the Camby/Wallace/Mutumbo, etc. dominance of the DPOY vote is another indicator.  But it seems like apples and oranges to me.  Are defensive rebounds part of playing defense?  Are offensive rebounds part of playing offense?  It seems to me that rebounding is its own category.  If you want to lump defensive rebounding into the discussion of top defenders, fine.  But it's only natural that it will skew the results of the DPOY vote further to the shot-blocking centers.

Nelson and Walker refer to the above three categories as the 'big three' of defensive statistics.  They go further and add four additional stats to the discussion:  

Personal Foul Efficiency - Measures ratio of steals plus blocks to personal fouls. Indicates how efficiently a player forces turnovers, removing the bias of jump-happy or slaptastic players who accumulate deceptive block or steal totals. Andris Biedrins' block total seems comparable to Ben Wallace's, until you realize that Wallace has more blocks with half the fouls. This statistic can be deceiving, since fouls may increase with defensive effort, but it is still a useful measure for comparative analysis.
Defensive Plus/Minus - Measures the difference per 100 possessions in points allowed with a player on the court versus off the court. The accuracy of this measure varies dependent upon how often you are on the floor and whom you share it with. This statistic is sometimes problematic, but in most cases it provides a good indication of a player's overall defensive value to his team .
Offensive Conversion Rate - Measures the percentage of a player's steals that result in a made basket or free throw attempts within 5 seconds. It is a measure of a player's transition defense in terms of how frequently his steals lead to offensive production. For example, during a game against Detroit, Ben Wallace anticipated a pass and stole the ball at the defensive end. He dribbled slowly downcourt and passed to Tyrus Thomas who was fouled on a dunk attempt. Wallace's anticipation led directly to offensive opportunity, giving the steal more value. In this case, the action took 6 seconds. Had the steal occurred in transition or facing up with his defender near the perimeter, he easily would have converted within 5 seconds.
Block Value - Measures PPG average of all opponents blocked. In most cases, a block against a good scorer is of greater value than a block against an NBA 12th man. Thus, each block has value relative to the player whose shot is blocked. A higher BV is typically an indication that a player is getting more quality blocks and/or drawing more difficult assignments. For instance, Orlando players Darko Milicic and Dwight Howard have roughly equal block totals, but if Milicic registered all his blocks against Michael Doleac, and Howard registered all his blocks against Dwayne Wade, the value of Howard's blocks would far exceed those of Milicic.

Without being too dismissive, let me just say... puh-leeze.  Foul efficiency?  Isn't taking a foul in the right situation (say fouling Shaq and putting him on the line instead of allowing him a dunk) a good and smart defensive play?  Foul trouble may impact the amount of time a player can stay on the floor and thereby reduce their ability to put up statistics in other categories, but to double penalize a defender for somehow not being efficient in their foul usage is statistical analysis on Spring Break - geeks gone wild.  I mean, I'm not even convinced that steals is such a hot measure of defense, so offensive conversion rate is only adding dubious value to a flawed measure.  Likewise for block value, the additional clarity it supposedly adds to the discussion is minimal.  Besides, in their own example, isn't the type of shot (essentially unknowable from game logs) far more important than the shooter?  Michael Doleac may be a doofus, but he can usually make a layup, and if Darko blocks that layup, it's still an important play.  

Defensive Plus/Minus on the other hand, I could get behind.  Of course all plus/minus ratios are problematic.  Player A represents only 1 out of 10 players on the floor at any point in time, so it goes without saying that the other 9 players have more impact on player A's plus/minus than player A himself.  Still, over time, it would seem that a reliable trend might emerge.  And, especially when discussing defense, where so many intangibles are involved (like switching, staying in front of your man, rotating, etc.), the plus/minus is the only stat that even attempts to get at the bottom line, which is whether or not the opponent scores.  It is also interesting to note that the top players in the league in defensive plus/minus ((Devin Harris, Bruce Bowen, Anthony Parker, Tim Duncan, Emeka Okafor, Jorge Garbajosa, David Lee, Rudy Gay, Andris Biedrins, Manu Ginobili) are, with the exception of Rudy Gay, all players that I do consider solid defenders, and many of them (like Devin Harris and Anthony Parker) underappreciated.  

So basically, I'm saying that I don't like any of this stuff, with the possible exception of defensive plus/minus.  Which leaves the question, how do you measure a good defender?  What constitutes good defense?

The answer is the same one that Supreme Court Justice Potter Stewart used when attempting to define obscenity:  I know it when I see it.  And I know that available statistical measurements are inadequate.

For example, Shane Battier, clearly a terrific defender, is not in the top in the league in any of these categories.  His defensive plus/minus of -3.7 is very solid, but he's not in the top 50 in steals or rebounding, and he was 45th in blocked shots.  Part of the problem of course is that he's not 7 feet tall.  But just as important is the way that individual statistical categories continue to be skewed by the pace of the game.  Houston's walk it up, paint by numbers pace simply makes it harder to put up any kind of statistics.  Maybe someday all stats will be per possession, but we're clearly not there yet.  

Interestingly, in going to such lengths to find new statistical measures, Nelson and Walker ignored one right there on the home page of drawn charges.  I could make many of the same arguments against the drawn charge that I do against blocked shots, but the bottom line is that it is in fact a MORE effective defensive play than either the blocked shot or the steal - the drawn charge ends an offensive threat, guarantees your team possession, and also adds a personal foul to an opponent, frequently a top scorer.  And what do you know, there is Mr. Battier, 9th in the league at taking charges.  (Anderson Verajao, another underappreciated defender, is number 1.)

The big problem with the statistics driven approach is that it skews the discussion to the shot blocker / rebounders, while NBA offenses are becoming more and more perimeter oriented.  True, a shot blocker as a last line of defense is a significant defensive asset, but in a league where Kobe Bryant, Carmelo Anthony, Gilbert Arenas, LeBron James, Ray Allen, Allen Iverson, Vince Carter, Joe Johnson, Tracy McGrady and Dirk Nowitzki are the top 10 scorers, perimeter defense is every bit as important if not more so.

As always, it's easier to criticize the choices of others than it is to justify your own choices.  Marcus Camby seems like a fine choice for DPOY, until you look at his defensive plus/minus.  Denver's opponents actually scored 2.3 MORE points when Camby was on the floor than when he was off the floor.  Like I said before, I know that plus/minus ratios have issues, but shouldn't the DPOY be so very good that his team's defense is measurably better when he's on the court?  

But my most vehement criticism, as always, is reserved for Kobe Bryant.  Or in this case, for the coaches who voted him onto First Team All Defense this season.  This selection is so head-scratchingly bizarre that I actually have to question whether or not my own bias is clouding my judgment.  But Kobe Bryant?

First of all, I find it strange that the performance of the team figures so prominently in the voting for things like All Star Games, All NBA and MVP, but seems to be completely ignored in defensive categories.  This is all the more strange when you consider that individual defense is so difficult to quantify, as we've already discussed at length.  Why isn't the discussion of top defenders limited to players on top defensive teams, in the same manner that All NBA candidates come almost exclusively from teams with winning records?  Anyone who watched even a single Laker game this season knows that they were among the worst defensive teams in the league.

Now, you might argue that Raja Bell of Phoenix and Marcus Camby of Denver might be similarly eliminated from the discussion.  However, although Phoenix and Denver do allow a lot of points, much of that is due to the pace of their games.  When measured by points per possession, Denver was the 11th best defense in the league, while Phoenix was 14th.   Not among the leaders, but at least in the top half of the league.  The Lakers on the other hand were 25th.  

There's also a common sense factor that seems to be ignored if Kobe Bryant is First Team All Defense.  Who does he guard?  You know who a coach considers his best defender by seeing how he matches up.  This point was obvious in game one of the Phoenix-San Antonio playoff series - Shawn Marion opened the game against Tony Parker, and opened the second half against Tim Duncan.  Meanwhile, Bruce Bowen matched up against Steve Nash.  Those are some tough defensive assignments given to some top defenders.

Contrast that with game 5 of the Suns-Lakers series, which occurred just after the All Defensive teams were announced.  First teamer Raja Bell opened up the game against the NBA's leading scorer Kobe Bryant.  Meanwhile, first teamer Kobe Bryant, opened up the game against... Raja Bell, the Suns 5th leading scorer and a player whose offensive repertoire is, shall we say, limited.  When the Suns brought Leandro Barbosa into the game with Nash, the Lakers countered with Maurice Evans, leaving Bryant on Bell.  Now, I'm not stupid - I understand that Nash and Barbosa are tough covers for Kobe, and I understand that Phil Jackson wants Kobe to conserve energy on the defensive end.  But you have to admit, there's something seriously screwy when a First Team All Defense guard is checking the Suns third or fourth best perimeter player while a 20 year old rookie is assigned to the league's two-time defending MVP and a journeyman is on the recently crowned 6th man of the year.  I mean, the key to stopping the Suns is containing Nash.  The Spurs put a First Team All Defense FORWARD on him.  The Lakers tried Jordan Farmar.  The Lakers would never dream of giving Bryant that assignment.  In fact common sense says that Kobe shouldn't be given that assignment, just as common sense says that he shouldn't be first team all defense (or second or third for that matter).

And it's not just Phoenix, an admittedly extreme example.  When was the last time Kobe took the toughest perimeter assignment on the opposition?  And I'm not talking about for a couple possessions down the stretch (although frankly, I don't remember many of those lately) - I'm talking for the game.  Now, Kobe apologists will say that he's CAPABLE of playing great defense, but that it's simply not his role on the Lakers.  As mentioned above, Phil wants Kobe's energies applied to scoring points, not stopping people.  That's great.  But in that case voting for him for the All Defense team because he COULD play defense if he chose to is like voting him into the Pro Bowl as a tight end.  He's 6'6", 240, has great hands, astounding body control, and I'm sure he clocks a pretty good time in the 40.  He could be a hell of a tight end if chose to.  But he doesn't play football.  

So he plays on one of the worst defensive teams and he doesn't take on the tough defensive assignments.  But the common sense argument goes even further - if you watch Kobe play defense, you know he's not good at it.  He ball watches, he gambles... basically, he plays defense like he's just waiting to play offense, and if the Lakers have to take the ball out of the net to get there, so be it.  

Like I said, I was beginning to think I was the only one who saw this, which troubled me.  I mean, even if the emperor is walking around naked, you do start to question your own sanity when everyone else says how nice his new suit is.  But then I came across a post in Roland Lazenby's LakerNoise blog.  In a conversation with Tex Winter comparing Kobe to Michael Jordan, Tex singled out Kobe's defense as being the area where he comes up short.

"I'd like to see him play better defense," Winter said, adding that he had addressed the issue recently with Bryant but didn't come away with the idea that Bryant was intent on changing his approach.
"You know Kobe," Winter said with a chuckle. "He has his game plan. I think he heard me. But he feels there's a certain way he's got to play the game. But it doesn't involve a lot of basically sound defense."
Because the Lakers need so much of his effort at the offensive end, Bryant has adopted a save-energy plan on the defensive end, Winter said. "He's basically playing a lot of one-man zone. He's doing a lot of switching, zoning up, trying to come up with the interception.
"The way Kobe plays defensively affects the team," Winter added. "Anybody that doesn't play consistently good defense hurts the team."

Aha!  I'm not crazy.  The emperor has no clothes!  No less an authority than Tex Winter sees pretty much EXACTLY what I see.  Not only is Kobe not playing GREAT defense - he's not even playing decent defense.  

Which leaves the question, why did 14 out of 29 NBA head coaches (Phil Jackson was not allowed to vote for him) vote him onto first team All Defense?  I have no answer to this question.  Plenty of basketball fans can't distinguish between good defense and bad defense, but I'm not arrogant enough to say that NBA coaches don't get it.  Interestingly, the other guards receiving votes (Raja Bell, Jason Kidd, Kirk Hinrich, Chauncey Billups, Devin Harris) are good defenders.  Why the blind spot with Bryant?  

The best answer I can muster is sloth.  Like the NCAA basketball coach who turned in a ballot every week with Temple ranked in the top 25 regardless of what Temple was doing, I can only assume that the coaches didn't want to put any more effort into this process than they had to.  In the absence of easily identifiable statistics that would point them towards another candidate (particularly hard to come by for guards), they just wrote down the name of a famous guard.  In last year's Awards Screed on the old blog (this seems like it's going to be an annual thing) I surmised that eventually LeBron and DWade would be on the All Defensive teams for similar 'path of least resistance' reasons.  I forgot that LeBron plays forward, and therefore will have to compete with shot blockers.  But lo and behold, Wade, who is far from a great defender, got a couple votes this season.  Coaches should know better, but apparently they don't.  

I wish I had a solution for all of this.  I think it's great that the NBA attempts to recognize defenders with these awards, since they are otherwise somewhat anonymous.  But short of having all of the coaches and voting media read my blog or call me asking for advice, I'm not sure how to fix it.  But a little common sense certainly wouldn't hurt.