Sports and Causality
The illustrious Malcolm Gladwell has a great article in this week's New Yorker dealing with the perception vs. the reality of greatness, particularly in sports. You can find it here. The article talks about Allen Iverson and his remarkable basketball skills. Then it talks about how two avant-garde basketball statisticians, through many man-hours of plugging and chugging, have made a startling revelation. They state that, although his statistics are good, Iverson does little to contribute to his team's success. They even ranked him as the 91st best player in the league in '00-'01, the year in which he won the MVP. Ouch.
This goes against everything that basketball fans see and think. They know he puts up 30 a game and has remarkable quickness and drive to the basket. And while everyone who watches the NBA will admit that Iverson is selfish and isn't the best shooter, they certainly wouldn't rank him 91st in the league in his MVP season.
A similar crisis of statistics is happening in baseball. The traditional measures of player efficiency--batting average, home runs, runs batted in--are being replaced by OPS (on-base percentage plus slugging percentage). Oakland A's general manager Billy Beane has made a career out of taking players with strong non-traditional stats and molding them into successful teams year after year. He manages to do this with a budget that is less than half of what the Yankees spend.
So what's going on here? I think the central problem of sports and statistics is causality. Things in life simply don't connect in cause-effect chains. They are messy and unpredictable. Mathematicians call this acute sensitivity to initial conditions, and it is explained in Chaos Theory.
Chaos Theory began when a meteorologist at MIT named Ed Lorenz made a computer program to simulate weather patterns. He would input some initial conditions, such as wind speed and temperature, and the computer would use basic meteorological principles to output new wind directions, speeds, and temperatures. Lorenz's machine became a fun guessing game around the office, as the MIT meteorology department would come in and place bets on which way the winds would blow.
One day Lorenz decided to re-run a series, so he re-inputed some parameters from a test the day before. To his surprise he got different results. Lorenz struggled to explain why he inputed the same numbers two days in a row and got different results each day, but eventually he figured it out. The second day's numbers were rounded off. Lorenz had assumed that a tiny, infinitesmal change to the original parameters would make no difference to the end result, but the opposite was true. Tiny, imperceptible changes in initial conditions make huge differences in results.
Nowhere do we see this more than in team sports. How can we quantify the value of a player like Iverson? We can't know, because we can't possibly know how much he contributed to the performance of a team on a given night. Chaos Theory tells us that the parquet of the floor or what the team had for dinner could have had a bigger effect. Would the 76ers have won more games if, for instance, Kobe Bryant or LeBron James were their star guard? No one knows, and conjecturing is dangerous.
That is why it is important to recognize the vast complexity of any complex system, be it a basetball team, a company, or the economy. The economy serves as an excellent example of the dangers of causality. I have heard people give credit to everyone from Ronald Reagan to George Bush, Sr., to Bill Clinton for the booming economy during the 90s. But the economy simply doesn't work like that. The boom-times of the 90s were accreditable to millions of factors, from the rise of day-trading to the spread of the internet. Assigning causality has become a tool for anyone who wants to make a point and needs numbers to back them up.
Thus Republicans credit the great economy in the 90s to Bush and Reagan, and the Democrats stick to Clinton. Likewise, Steven Leavitt, economist extraordinare, attributes the precipitous drop in crime in New York to abortion--a conclusion that I doubt a pro-life economist would have come up with. And likewise we have to realize that people who rank Iverson low probably have an agenda, too. I suspect that they miss the good old days of basketball, when shorts were short and players were unselfish. We need to realize that numbers are just as subject to bias as anything else.
Fear the Greeks, even when they bring numbers.
This goes against everything that basketball fans see and think. They know he puts up 30 a game and has remarkable quickness and drive to the basket. And while everyone who watches the NBA will admit that Iverson is selfish and isn't the best shooter, they certainly wouldn't rank him 91st in the league in his MVP season.
A similar crisis of statistics is happening in baseball. The traditional measures of player efficiency--batting average, home runs, runs batted in--are being replaced by OPS (on-base percentage plus slugging percentage). Oakland A's general manager Billy Beane has made a career out of taking players with strong non-traditional stats and molding them into successful teams year after year. He manages to do this with a budget that is less than half of what the Yankees spend.
So what's going on here? I think the central problem of sports and statistics is causality. Things in life simply don't connect in cause-effect chains. They are messy and unpredictable. Mathematicians call this acute sensitivity to initial conditions, and it is explained in Chaos Theory.
Chaos Theory began when a meteorologist at MIT named Ed Lorenz made a computer program to simulate weather patterns. He would input some initial conditions, such as wind speed and temperature, and the computer would use basic meteorological principles to output new wind directions, speeds, and temperatures. Lorenz's machine became a fun guessing game around the office, as the MIT meteorology department would come in and place bets on which way the winds would blow.
One day Lorenz decided to re-run a series, so he re-inputed some parameters from a test the day before. To his surprise he got different results. Lorenz struggled to explain why he inputed the same numbers two days in a row and got different results each day, but eventually he figured it out. The second day's numbers were rounded off. Lorenz had assumed that a tiny, infinitesmal change to the original parameters would make no difference to the end result, but the opposite was true. Tiny, imperceptible changes in initial conditions make huge differences in results.
Nowhere do we see this more than in team sports. How can we quantify the value of a player like Iverson? We can't know, because we can't possibly know how much he contributed to the performance of a team on a given night. Chaos Theory tells us that the parquet of the floor or what the team had for dinner could have had a bigger effect. Would the 76ers have won more games if, for instance, Kobe Bryant or LeBron James were their star guard? No one knows, and conjecturing is dangerous.
That is why it is important to recognize the vast complexity of any complex system, be it a basetball team, a company, or the economy. The economy serves as an excellent example of the dangers of causality. I have heard people give credit to everyone from Ronald Reagan to George Bush, Sr., to Bill Clinton for the booming economy during the 90s. But the economy simply doesn't work like that. The boom-times of the 90s were accreditable to millions of factors, from the rise of day-trading to the spread of the internet. Assigning causality has become a tool for anyone who wants to make a point and needs numbers to back them up.
Thus Republicans credit the great economy in the 90s to Bush and Reagan, and the Democrats stick to Clinton. Likewise, Steven Leavitt, economist extraordinare, attributes the precipitous drop in crime in New York to abortion--a conclusion that I doubt a pro-life economist would have come up with. And likewise we have to realize that people who rank Iverson low probably have an agenda, too. I suspect that they miss the good old days of basketball, when shorts were short and players were unselfish. We need to realize that numbers are just as subject to bias as anything else.
Fear the Greeks, even when they bring numbers.
posted by Tim Dees at 1:52 AM
0 Comments:
Post a Comment
<< Home