Wednesday, May 13, 2009

The Fall of the American Hoops Empire

This is a bit late, but I'm finally getting back to some posts that I had started a few weeks ago, but never finished.

(This post was originally started on 3/20/09)

I hate to sound like an old fogey, but really, if you think about the logic that prevails when it comes to modern, big-time basketball, you start to see some of the issues that permeate the game. Take a look at this NYT Mag article on 13-year old phenom Allonzo Trier.

Now, I don't know how this kid's going to pan out. There's a legit chance that he'll end up being a really good player, and will go to college or Europe for one year before the NBA (with it's silly age limit). And he may even be the next Cousy, Stockton, Chris Paul, J-Kidd, or D-Will. Of course, he may end up being the next Sebastian Telfair, Kwame Brown (really not so bad, since they made the NBA, and made a few million along the way), or Ndudi Ebi, Korleone Young, or Lenny Cooke (guys who never got drafted or made rosters).

But what I'm more concerned about is the prevalence of the AAU, club-level basketball, and what it's doing to/for American hoops. On the bright side, you teach some competitiveness and you give kids the chance to play year-round. But there's some concerns about the potentially shady things that are going on (money under the table, kids being guided to sham high school degrees, etc.). And one major issue is that the coaching is probably shaky, at best. But that's not the thing that really concerns me. What gets me worried is that the lessons, at the broad level, that are taught are just not conducive to winning basketball. Certainly, there are X's and O's that are probably not there in AAU games. But the scary thing is that the reward system is set up so that everyone's gunning for individual accolades.

People are remarkably good at adjusting to their surroundings, and maximizing their utility. If you put people in a system where they're rewarded for individual performance, they'll devote their energies to standing out. Now, normally, the way to adjust for this is to figure out a better reward system. But people are generally bad at figuring out the bigger picture, and prefer to rely on information that's easily observable. And what we can easily observe are the actions that become sports stats.

Basketball statistics are an evolving beast. The game is free-flowing, and it's hard to record all of the events that are happening. Players move all over the place. And the stats that are generally reported are things like points, rebounds, and assists. These aren't necessarily good at predicting how well a team will do, or for that matter what a player's contribution is to winning. And when our stats don't do a good job of predicting success, you run into problems - when the most salient things are misleading, you're likely to just get your self more lost in the forest.

About 30 years ago, in baseball, a bit of a statistical revolution started (cf., James, Bill), which really started picking up momentum in the mid-90's when the internet became a household utility and fantasy baseball started to go mainstream. People were able to come up with more and more refined statistics that have helped redefine how the game is played at the top level. A key factor in this was the nature of baseball. Baseball had a number of discrete events that lended itself to analysis. And there are pretty discrete zones that correspond to where defenders play. Plus, people had been keeping baseball stats for ages - one of the things that dads do is teach their kids how to keep score at a baseball game. The net result, is that we've see an evolution of the stats that everyone knows (20 years, everyone was talking about batting average and ERA, and they still do, but a lot of fans know also about OBP and WHIP). These better stats have helped in the evalution of baseball talent.

In basketball, when we try to figure out who the best players are, our best guess is that people who score a lot of points, rebound the ball, dish out assists, are the best players. And for the most part, we can see how this bears out with superstars. And at the AAU level, a talented kid can be a relative superstar, and wants to be a superstar. So, what do players do? They gun for those stats. But what happens to players that aren't superstars? And more to my concern, what happens to teams that have players who aren't superstars trying to be superstars?

Well, you get some obvious issues that we've seen in the past. Teams sign guys to max-contracts, and expect them to become legit alpha dogs out there. But there are only so many Kobe, LeBron, Hakeem, or MJ-level players out there who can single-handedly will a team to victory on a regular basis. Look at the top 50 players in NBA history. Even on that elite list, if you look at the modern players (I have no insight into the old timers), I'd say that you have only a few true top-flight guys (MJ, Magic, Bird, Hakeem, Barkley, Isiah) that were true top-dogs. Even guys like David Robinson, Scottie Pippen, James Worthy, Shaq, etc. weren't true top-dogs. Most top players aren't legit alpha dogs. And teams make mistakes when they try to see if people grow into that role. Teams are starting to become smarter about it, but you still see stuff like the Rashard Lewis contract or what Jerry Stackhouse was signed to a couple years ago, where teams pay way too much for a guys who is a nice 2nd or 3rd piece to the puzzle, and then expect the guy to be the next Kobe or MJ. Those teams tend to end up being regular lottery teams, until they jettison the mistake and rebuild (cf., Allan Houston, who was one of my favorites, but even I knew that he wasn't the guy even if he was healthy - and then they made a rule up named after him for teams to drop their mistakes from their salary cap!).

What a bit more problematic is what teams do with their roster, with regard to role players. When players are asked to do things that help the team win, but may not correspond to how the general populace evaluates players, they often balk. And this makes sense. Players play for their next contract. And when the rest of the NBA evaluates players using traditional standards, the players know that they should put up traditional numbers so that the rest of the NBA will be more inclined to sign them. Shane Battier and Daryl Morey of the Houston Rockets are one of the the exceptions to this. But part of that is the makeup of Battier (played at college in a very team-oriented system), and the vision of Morey (a stat guy, dubbed "Dork Elvis" by the Sports Guy).

In baseball, teams can develop an organizational strategy that permeates through it's farm system. A while back the A's had certain rules. If you didn't draw enough walks, they wouldn't promote you from A to AA. If you didn't have a high enough OBP, the wouldn't promote you. So, you saw a culture built. And, at the major league level, they signed guys who fit into their offensive strategy. The incubation period player of a baseball player is often longer than that of a basketball player. Baseball players typically advance through the minors, hit the majors around 23 or 24, and those who stick tend to improve over the next 4 or 5 years, plateau for a few years, and then decline starting at 33, and generally are out of the game by 35. And some baseball players come out of nowhere and become stars (Mike Piazza is the prime recent example there). Basketball players seem to come in from day 1, and either have "it" or not. And you don't see as many basketball players coming out of nowhere to become hall of fame level players. They'll become role players or even occasional all-stars, but generally not true alpha dogs.

So back to AAU ball. What lessons are taken away? Everyone tries to be the superstar because everyone's trying to get theirs. And as a result, you get a whole bunch of candidates to be the next alpha dog. The problem is that a lot of people don't have the psychological makeup to be Shane Battier, when it becomes apparent that they aren't the next Kobe. But they're so good, so athletic, so full of potential to be great 2nd options or great role players, and teams draft them. But then the clash starts. Players who have always had the ball in their hands, falter when they have to play off the ball. Players who have been able to get away with just being there on defense, balk when they need to lock up their guards. Players who are used to shooting 25 times a game get offended when they only shoot 8 times a game, often with no regard to whether or not they should shoot more than 8 times a game.

Watch the Atlanta Hawks play in the half court. Or Rasheed Wallace when he doesn't want to play in the post. That's ugly basketball. Or worse, watch those streetball shows on ESPN. That's not even basketball. That's basketball that happens when there's not an effective system in place or when players ignore the system. You see this all the time at the gym in pick-up games. This is what a lot of AAU games look like, and this is what you see want to trickle up into college and the pros. In college, coaches are still kings or dictators, and can rule with an iron fist. But in the pros, for the most part the players run the asylum. And if you set the incentives such that individual performance on relatively tangential (at least in regard to winning) stats results in the reward, you end up with a mess. And having this exist over a long period of time (i.e., the AAU system) only exacerbates the problem. And if you listen to the players on the streetball shows on TV, you hear tinges of bitterness, where the self-view is so high, but so incompatible with a winning system so as to prevent the NBA dream. And you wonder if the system hasn't failed a lot of people, at the benefit of a few powerbrokers and a few of the superstars that probably would have been superstars anyway.

We'll see how Allonzo Trier develops, how much he grows. If he stops growing at 5'7", we probably don't have to worry about this, since he'll be too small. If he grows to be 6'5", we probably don't have to worry about this, since he'll be a bigger Derrick Rose and will be so great that it won't matter. And there's always the chance that he and his immediate support system are going to be smart/strong enough to withstand the potential pitfalls of a skewed system, and still develop into the star that people think that he is destined to be.

-Chairman

2 comments:

Westy said...

Good post, Chairman. These are certainly areas of much interest to me. Funny you should post this right after I did my latest post on basketball recruiting.

I totally agree that the incentive structure based on especially points scored starting at a young age does not do a good job of developing players who could be the next great Battier or Rodman, and thus the best future teams for the USA and NBA. Not sure how to change it though, because even when we develop reliable more advanced stats at the NBA level, points are still the easiest to track in junior high.

Continued discussion of more advanced stats is good, though. I sure hope you're paying attention to the conversation at the APBRmetrics forum. Maybe next year we can go to the Sloan Conference at MIT together.

I would note though that, based on your own standards of valuing the right stats, your inclusion of Isiah Thomas (and probably Barkley too) amongst the "top-flight" players is probably in error. As Basketball-Reference (really like the SPM system he's developed; think it's one of the most accurate out there right now) does a good job of showing, he is likely overrated.

Chairman said...

The Isiah inclusion (despite my disliking him very much) is because he was the alpha dog on a championship team. Similarly, Barkley was the lead dog on one of the best teams to never win a championship (and had the swagger - Ewing, who was one of my faves, was never really the lead dog, which allowed Starks to shoot my Knicks out of a championship).

I took a peek at the SPM stuff.

http://www.82games.com/comm30.htm

Obviously this stuff is useful, and probably better than what's out there now. But intuitively, I'm not fond of the reliance of a relatively macro-level "outcome" stat, which is what the use of +/- is, since the scientist in me wants to look at the first-principles (i.e., break things down to the essential core). Basically, what I don't like is that at it's most fine-grained, this method can take you down to the possession (and should be about on par with refined scoring efficiency stats).

Where I want to see things go is down further to the components of a possession. Of course doing so with existing stats has the problem of requiring that someone has to get assigned credit for every mini-outcome.

Just with regard to moving the rock, I want to see stats on how effective a pass is (i.e., does it set up a teammate to dribble, drive, or pass). I want to see how often people throw challenged passes and how often they succeed. I want to see how many passes per possession for when teams score and for when teams don't. I want to see how often players miss passes to open teammates.

Somehow, we can capture all of this very intuitively in our minds in a split second, though we have inherent biases. Where a good model can help is in filtering out the biases. However, if we want to build predictive models, I still think we need to get to data that is appropriate.