Thursday, April 02, 2009

Statistically Useless

One of the nice things about living in the Bay Area is that we have a lot of major professional sports teams: two NFL teams, two MLB teams, one NBA team, and one NHL team. Never you mind that they all stink currently, except for the Sharks. With so many teams, sports radio is a lot of fun. Today I was listening to Fitz & Brooks as they fielded a question from a listener about the Golden State Warriors and their lack of defense. They made the case that the Warriors have a lot of defensive talent because they are 6th in the league in steals and 1st in the league in blocked shots. They argued that the Warriors just needed to allow less "blow-bys." I looked around, but could not find this statistic.

I was almost really impressed by Fitz & Brooks. It seemed like they were on the verge of saying "NBA statistics are meaningless" but they could not go that far. The NBA keeps a lot of statistics, but they really are mostly meaningless. Points, rebounds, assists, steals, blocks, etc. None of these are particularly good at determining if a player will make his team more or less likely to win a game. That is why you see guys like David Lee, Troy Murphy, Jose Calderon, and Mario Chalmers as statistical leaders. Look at the league leaders in blocks. Their teams are ranked 8th, 18th, 25th, 30th, and and 20th in the league in points per game. In terms of team field goal% allowed, the ranks are 3rd, 4th, 25th, 23rd, and 19th.

The NBA is hardly the exception. Most sports have a lot of meaningless (from the standpoint of predicting team success) statistics. Baseball has terrible stats like batting average, runs batted in, stolen bases, and fielding percentage. Football is not quite as bad mostly just because there are less stats. Still things like rushing touchdowns and even passing yards are pretty worthless.

Of course baseball has seen a statistical renaissance and it is known as sabremetrics. Michael Lewis's Moneyball is a popular work that discusses the value of sabremetrics. It uses the success of the Oakland A's and their GM Billy Beane as proof, but if it was written today, then the success of the Boston Red Sox would be (more?) compelling. Lewis expanded this "nerdy numbers leading to unexpected success" plot to basketball recently by writing an article about the Houston Rockets' Shane Battier. Basketball is tougher sport to measure though. Maybe there will be sabremetrics in basketball eventually, who knows.

It was definitely easier in baseball, where some of the statistical ingredients for more meaningful sabremetrics were already measured (like number of walks, total bases, etc.) What would similar stats in basketball be? Certainly making shots is good, missing shots is bad. How do you measure a three point play vs. a three point field goal? If a player takes a shot and misses, but somebody else gets an offensive rebound and an easy field goal, how is this measured?

No comments: