I think over the next decade we're in for a huge shortage of analysts and statisticians in almost all fields of life. Availability of data is ever increasing. The benefits from using it more effectively than competitors are immense. In almost all of the startup businesses in our portfolio (venture capital firm), we've now hired number-crunching guys who do nothing but metrics and we're seeing the results.
shortage? nah. now those finance majors who were making millions on juggling mutual funds can get a real job and contribute to society rather than living off unearned commissions. (if your mutual fund does worse than the economy as a whole you don't deserve reward, you are worse than random input)
First things came to my mind being that schools would use these data to determine admission. What kind of applicants will yield the best pay off? (endowment?) Same for companies hiring employees.
Interesting idea. One of the things I found interesting about Moneyball is that the Oakland A's were willing to pay premiums to role players, and older (presumably on their way down) veterans.
Michael Lewis is a non-fiction author. Most relevant to this thread, his name is instantly recognizable among people who are interested in the intersection of baseball and statistics, because he wrote Moneyball. The basic idea is that historically the market for talent in baseball has severely mispriced talent because it overpays for attributes that do not win ballgames. Thus, if one knew what actually won ballgames, one could thwomp the competition while spending much less money. And, sure enough, someone did just that.
For the rest of the story, I recommend the book. I have no interest in sports whatsoever but one of my stats geek friends does, and he accords it a place of honor in his bookshelf.
"thwomp"? The Rockets are a good team but haven't won a championship and haven't been that competitive in the playoffs.
The "moneyball" idea doesn't translate quite as well in basketball. One, it's a significantly less statistical game than baseball. Athleticism doesn't translate perfectly to numbers and individual stats can be largely effected by teammates.
That the linked article is by Michael Lewis. A very important data point, since I would probably read Michael Lewis if he just reprinted the phone directory. He would find some way to make those numbers compelling.
Great article. As a basketball player in my youth, I always struggled with the dynamic of selfishness and selflessness inherent in the game and this was best attempt I've seen at explaining it. There really are so many way a player can help his team that do not improve, and even decrease, his stats. The use of statistical methods to identify this, like Moneyball did for baseball, is fascinating.
Fun article. I loved Moneyball, and enjoyed other stuff Michael Lewis has written. I'm glad he's looking at basketball, since that's really the only sport I actually follow a bit, mostly through a blog dedicated to sports stats that happens to focus mostly on basketball: http://dberri.wordpress.com/.
The author (Dave Berri) is an economist who's done some work on connecting team success to the actions of individual players through stats. A major theme is that many players try to maximize their paydays by over-emphasizing scoring, and that this is actually reasonably sensible on their part because decision makers in basketball currently overvalue scoring, and pay accordingly. It does look like that's slowly changing, at least according to this article.
the player that is not a star but helps his team wins championships is James Posey. Every team needs a guy that does the dirt work and nobody is more effective at doing that right now than Posey. Coming off of the bench he played a huge part in both Miami Heat and Celtics championships. He plays great defense and is also an offensive threat from the outside.
By that logic, every team that wins championships also needs at least two superstars. The Heat had two, the Celtics had 3.
While I love Posey (Celtics fan here), it takes a foundation of superstar talent first to even have a chance.
Many people take for granted the 20-30 points, that the team scoring leader puts up every night, and start to look elsewhere for that magic winning element. Without the consistent play of the scorer (and the attention that draws from defense) guys like Posey would be exploited by a good team.
I think the point of the article is that straight statistics like that are not always accurate. For example, what if a player allows the person he is guarding to score more points on average, but during the time is he is on floor, the opposing team tends to score fewer points. This is easily possible if the player is a great help defender who tends to stray from his own man to help his teammates.
This sounds good in theory but there are a TON of variables that would be really hard to encapsulate. I'm not saying it can't be done, but (good) defense in basketball is extremely team-oriented. Positioning, rotations, double teams (and the rotations that accompany them), etc., not to mention zone defense.
To give a more concrete example, a team with a great defensive center gives the guards flexibility to play tighter man to man, because they can more readily afford to let the guy they're guarding get by them (thanks to their "backup", the center). This has the additional consequence of implying that they give up less room on jump shots.
Another thing to consider is steals. A steal is really big because it obviously prevents the opponent from scoring, and frequently leads to fast break opportunities. However, it's misleading to simply look at the number of steals a player gets. Some players get steals within the "flow of the game", whereas others get them through gambling (I don't have a quantitative source for this, but anyone who's watched Larry Hughes play knows what I'm talking about) for steals, meaning that whenever they fail to get the steal, they also end up way out of position, often leading to an easy score for the opponent.
says Nancy Grasmick, Maryland's state superintendent of schools, in a recent article in the American School Boards Journal. "If you're teaching to the test [...] then we don't see anything wrong with that."
Read it a few hours ago and really enjoyed it. The recognition of all of these teams that there are ways to build a team beyond the conventional is smart thinking, even if doesn't always result in titles out of the box..
