Thursday, November 13, 2008

Questions about the Answer

Got a question recently:

Mr. Fruithoopz-
Loyal reader. Wanted to know if you could answer a question. Lots of sportzwriterz and fans like to accuse Allen Iverson of being a ball hog. As a student of the game and the numbers that make
them, is it accurate to refer to Mr. iverson as a ball hog? Is there such thing as a good ball hog vs bad ball hog? I know it is has a bad connotation in the sports pages. Is it true or is it just some people don't
care for his personality and perceived impact on the game and use this term pejoratively? thank you!

I should start out by saying that this post might be more biased than usual because I just really like AI. That said, let's take a look.

I. The Color Commentator dictionary
Basketball commentators like to use contextual and often suggestive vocabulary to describe playing style. For instance, "intangibles" and "hustle" and "energy guy" are words associated with rebounders. In a similar vein, "unselfishness" is generally measured by assist totals.

"Ball Hog" is an interesting concept. Theoretically, it should connote the same thing as "selfish," but in fact players like Allen Iverson and Stephon Marbury, who have had high assist totals, can still be ball hogs. And it's not just ball-domination that makes a ball hog, as both Lebron James and Steve Nash have made careers out of dominating the basketball on offense and yet have avoided being labeled as ball hogs.

II. Value
Allen Iverson, as an extremely high usage but average efficiency player, is in the middle of an ongoing unsolved problem of individual basketball statistics -- that is, the question of the tradeoff between usage and efficiency. We've discussed here before that most players tend to lose some efficiency when they are forced to use more possessions on offense. How much a particular player's efficiency will drop off will depend on his particular skillset and the role he's asked to play in an offense. However, in general, the question of how to weigh the value of a player's usage contributions vs. his efficiency is still an open one.

The only reason the question is so important at the moment is because of the ESPN-ization of statistics. It is not enough, currently, for audiences to see numbers that show what a team does well and what it struggles with, and the roles of the various players on the team and how well they are fulfilling them. Instead, we are bombarded with MVP lists and lists of "best players ever" and so on and so forth. In that context, statistics are being asked to answer context-independent questions about player value, forcing us to make absurd and ultimately useless comparisons between players like Kobe Bryant and Kevin Garnett. With this need for all-encompassing player metrics, the question of the relative values of usage and efficiency become very important. How valuable is a created shot? When Steve Kerr hits a wide open three pointer, how much credit do we give him for making the shot, vs. how much credit we give Michael Jordan for allowing Kerr to be so open in the first place?

The two most prominent overall player metrics currently are Dave Berri's winscore and John Hollinger's PER. Berri's metric was developed by valuing boxscore statistics at the team level and then just applying those values to the individuals who contribute them. Since it is developed in that way, the metric is blind to individual usage contributions (at the team level, usage is not in doubt -- every possession gets used somehow, by definition), and as such, finds Allen Iverson to be an average player. PER, on the other hand, values usage highly, even if that usage is somewhat inefficient. As such, PER sees Iverson as one of the best players in the league.

Adjusted plus-minus, which tells us how a team performs when a player is in the game, tells us Iverson is somewhere in between the extremes -- one of the top offensive contributors from the 2-guard position, just behind transcendent talents like Kobe Bryant and Dwyane Wade.

(This is as good a time as any to mention that I'll be mostly ignoring the question of defense because, in short, Iverson has never been a very good defender. This is due partly to some effort lapses but moreso to the fact that he's just not big enough to guard most of his opponents).

Neither PER nor WinScore take into account supply and demand in their valuations of shot-creation. Just how rare is the ability to create quality scoring opportunities, relative to how rare it is to be able to consistently hit open shots when you receive the ball? The answer to that question ought to tell us more about Iverson's value to a team, vis-a-vis winning.

III. Innings Eaters and Economics
One of the underreported advantages that Iverson brings is his remarkable ability to stay on the floor. Over his career (spanning 835 games and counting), he has averaged 41.7 minutes per game. The only other players in history to have averaged over 40 minutes per game are Wilt Chamberlain, Oscar Robertson, Elgin Baylor, Bill Russell, and Lebron James (and James has only been at it for 399 games).

The basketball regular season is LONG. In baseball, due to the long regular season, we understand the value of an "innings eater" who can keep other pitchers rested. In basketball, such players are so rare that we don't even consider their value. But a player who can produce at average efficiency (which Iverson does) for almost the entire game gives a GM the opportunity to save money on a replacement-level backup, thus allowing him to spend more in another area. But Iverson doesn't just produce at average efficiency -- as discussed in part II, when Iverson is on the floor, he takes up enough possessions (through shots, assists, and turnovers) for over 1.5 average players -- you can put him out there with a couple of Ben Wallace-type role players without really hurting the offense, since Iverson is able account for so much of the offense. To use another baseball analogy -- imagine a decent hitter who is for some reason allowed to bat twice each time through a lineup (when no other team is allowed that advantage), once for himself and once in the place of any other batter on the team. Even though the hitter is only decent and not spectacular, he greatly betters the team by allowing them to start another defensive specialist without paying the price offensively.

IV. Making Players Better
There's a lot of talk about how a real star "makes players better." It's unclear what this means, and that's a good thing -- if there were a solid measure of such a quality we wouldn't talk about it so much. Let's imagine offensive scenarios in which a player does make his teammates better:

A) Gravity -- A player draws so much attention that he warps the defense, leading to wide open outside shots and cutting lanes to the basket, as well as easy offensive rebound opportunities for teammates.

B) Precise passes -- (not to be confused with just passing in general). A couple of examples: noticing immediately when a post player has established position and getting him the ball in a position to score, rather than waiting a second or two and then struggling to get him the ball, forcing the post player to step away from the key and then try to re-establish position to avoid getting a three-second violation. Or: getting the ball to an open shooter at just the right height so he can catch and shoot in one quick motion -- not passing below his waist or above his head.

C) Shouldering a disproportionate offensive burden -- This is where Iverson fits in. Most players in the NBA can be efficient on the offensive end as long as they have defined and limited roles. Spot up shooters can shoot high percentages as long as they can move without the ball, catch, and shoot, without having to put the ball on the floor. Finishers and rebounders can be efficient if they aren't given the ball outside of about 8 feet from the basket. For this reason, shot-creators (for instance -- Steve Nash, Kobe Bryant, Lebron James) are valuable since they allow role players not to stray from their roles. In Iverson's case, imagine a defense that stays at home on shooters, and pushes big men out of the paint. With Iverson around, the role players on the team aren't forced to take shots that they can't hit regularly (imagine Ben Wallace shooting an 18-foot jump shot), because Iverson will take the shot himself, with a much less dramatic dropoff in his ability to hit it (Iverson will hit an off-balance 12-foot runner much more frequently than Ben Wallace will hit that 18-foot set jump shot). Perhaps looking at an example will help -- let's look at Anthony Carter, who has played with Iverson in the backcourt in Denver. Below is a graph showing the relationship between Carter's usage and his efficiency:

Notice that Carter is at his most effective when using less than 15% of possessions. There are five players on the floor at any time, so an average player would use 20%. So for Carter to be effective, he needs to play alongside someone who can use an extra 5% of possessions. Enter Allen Iverson, who for his career has used 32.4% of possessions while he's on the floor (at the same efficiency as Carter when Carter uses around 12%) -- he can use the extra five to help Carter and still has enough offense in him to add another Carter-like role player (say, Kenyon Martin). These role players are then allowed to not only be effective in their offensive roles, but to fill defensive roles better than an average offensive contributor might (for instance - Kenyon Martin continues to be one of the better defensive players in the league, depsite his offensive shortcomings).

If Iverson is making players better in this way, we should be able to see some sort of evidence. In fact, we have player-pair data to show us that Carter did in fact shoot a better FG% and commit fewer turnovers when he played alongside Iverson (41.4% without him, and 46.4% with him). The same is true for several other role players (although, interestingly, not all of them). Last year's player pair data is hard to extract too many conclusions from just because Iverson played almost every single minute of the Nuggets' season, which is unusual (and, for a 32 year old, pretty impressive). So, here's the player-pair data from the year before, when he was only with the team for 50 games, and Philly's data for 2005-2006 (for instance, with a little arithmetic we see that Kyle Korver shot 33.3% without Iverson on the floor and 45.7% with him on the floor).

D) Drawing fouls -- One of the reasons Iverson is able to produce with even average efficiency despite low shooting percentages is his ability to constantly draw fouls and get to the free throw line (averaging 9.3 free throw attempts per game for his career -- the only other guards in NBA history to average over 9 for a career are Jerry West and Dwyane Wade). While obviously this helps the team score points, it also helps in other ways. With Iverson able to draw fouls early in quarters, he is able to get the opposing team into the penalty, which allows all of his teammates to take foul shots later in the quarter after minor hand-check fouls on the perimeter. Last year, Denver was second in the league at scoring from the free throw line, scoring close to 27 points from the line for every 100 shot attempts. They were first in the league by a huge margin in free throw attempts for the season. Another advantage of being able to draw fouls is getting the top players from the opposing team into foul trouble and forcing them either to the bench or to be hesitant on defense. There isn't currently a very good measure for this effect.

V. Detroit
So how does Iverson fit in in Detroit, a team which without him has consistently been one of the most efficient offenses in the league? I have no idea. I've said here before that I saw the trade as purely about building for the future with his expiring contract and turning the team over to Rodney Stuckey, but Iverson is still a remarkable player so he has to have some impact. We do know that Iverson is able to adjust his game to some extent -- last year in Denver he used a career low percentage of possessions (26.7%) and had his most efficient year ever (115 points per 100 possessions, which is well above average, in Kobe Bryant territory), shooting career highs of 48.8% eFG% and 56.7% TS%, while establishing a career low in turnovers per game. Now that he's with an even better offense, will he adjust even more, become more efficient? He's already discussing the changes:

"The game is a lot easier playing with these guys," Iverson said. "They're so unselfish. All they care about is winning. So many guys can do so many things, and they make it much easier for me."

So, let's wait and see. And if Iverson can have any sort of success in Detroit, then it's really time we got rid of this whole notion of selfish ball-hoggery.

VI. Aesthetics
We don't watch basketball just to see who's going to win or what someone's shooting percentage will be. Iverson matters also because we have to watch him. His crossover has almost become cliche over the last ten years, but the whole generation of crossover artists is following in his footsteps. And his ability to constantly, year after year, game after game, get to the rim and finish despite giving up 12 or more inches to his opponents in the paint is the sort of thing that makes me feel warm and fuzzy. The creativity involved in getting a ball over and through bigger players and into the hoop is constantly amazing. And Iverson matters because he's a cultural icon. He represents so much of an entire era of NBA basketball between Jordan and Lebron, of rule changes and image problems and changing skillsets and so forth. I really wonder if the correct historical comparison isn't Pete Maravich, another somewhat inefficient but aesthetically unique high-usage dribbler/passer/scorer who, like Iverson has at times in his career, went underappreciated. And this is probably where the question of "ball-hog" most belongs -- how do you as a basketball fan feel about this particular style of play? We've seen that it can be effective, so it's not a moralistic question about what "leads to winning" or anything, just -- is this something that is enjoyable to watch?


  1. thorough and fascinating! particularly the innings eater + economics section. where is that anthony carter chart from?

    the typical color commentator has also predominantly been skewed towards favoring the textbook pick&roll play of bball fundamentals and considering athletic-based juke-moves distasteful. AI has been at the forefront of shifting that perspective and discussion, even if most commentators are still struggling to adjust.

  2. Thanks!

    The Carter chart is something I created using Play-By-Play data and on-off data that I downloaded from The idea for that type of chart comes from the book Basketball on Paper by Dean Oliver, although I have flipped the vertical and horizontal axes. Oliver calls the charts "skill curves" and uses them to analyze the usage-efficiency relationship for individual players (it's been more generally looked at in studies like this one), specifically to determine ideal roles in an offense for different types of players. It becomes much easier to talk about how possessions ought to be distributed once you can see what contexts allow players to be most effective.

    There's definitely a lot to be said for the fundamentals, but every now and then, especially in the NBA, someone has such a unique skillset that you have to just enjoy it without the textbook. A lot of the color commentators do, I think, get that, but there are still those who aren't into anything but the classics . . ..