More Statistical Tomfoolery

Arcane statistics have created a sub-culture among sports fans. The geeks have a place in sports now, especially in baseball, and I'm wondering if across America the math club is gaining street cred in high schools by offering their analytical scouting services to the jocks. The most recent example of a statistical analysis to backup common sense, and disprove its methodology with shoddy results can be seen over at The Hardball Times.

The "geek" in question sought out to use statistical analysis to find the "Best Outfield Arms" of 2006. He used 5 situations as the basis for his calculations:

1. Single with runner on first base (second base unoccupied).
2. Double with runner on first base.
3. Single with runner on second base.
4. Fly out with runner on third base, fewer than two outs.
5. Fly out with runner on second base, fewer than two outs (third base unoccupied).

Then he compared how outfielders fared in each situation. Did the runner successfully advance, did he not even try to take the extra base, or was he thrown out trying to advance.

OK, I'm with him so far. If you were to never actually watch a game of baseball, this might be the best way to judge an outfielder's arm. And his results, on first glance, look like they reflect what common sense would tell us. Alexis Rios, despite his name, does not throw like a girl. Johnny Damon does. Abreu is middle-of-the-pack, Melky is near the top.

Upon closer inspection, though, the logic breaks down. For example, Manny Ramirez is ranked higher among left fielders than Melky Cabrera. OK, anyone that's ever watched the two play realizes that Melky has a cannon and Manny has a peashooter. There are two reasons why this flawed statistical system spat out Manny's name higher than Melky's.

Both reasons stem from the fact that holding runners is heavily weighted within the statistical set. So, when a runner is on first and does not even try to advance to third on a single, that's a positive for the player. In Manny's case, he played half of his games about 150 feet away from the plate in front of the green monster. The dimensions of the field hold runners, not Manny's arm. Conversely, Melky played half of his games in cavernous Yankee stadium. He has to cover about 3 times as much ground as Manny, so the field is responsible for NOT holding runners to some extent.

The writer admits some shortcomings of his analysis at the bottom of the article, though not all of them. I think this is an example of how you can't replace good old scouting for some areas of the game, as much as Theo Epstein and Billy Beane want you to think otherwise.