Friday, November 19, 2010

In defense of the statheads

As you know, Felix Hernandez, he of the 13-12 record, won the A.L. Cy Young over, among others, CC Sabathia. The award was announced yesterday. Twitter was a flitter all afternoon with the announcement of a sea change in the awarding of awards. Praise (or laments, depending on your point of view) for the process overshaddowed Hernandez's actual award.

I had to travel about six hours in the car today, which meant listening to the likes of Kevin Kennedy discuss how the win has been devalued by the sabermetric community. Nothing could be further from the truth.

I have never heard a stathead say that a win is bad. In fact, in my experience, statheads value only the win. They spend considerable time and effort, and a lot of brain power, sifting through data with one goal in mind: trying to determine how teams win. The idea (and someone can correct me if I'm wrong) is to crunch as much data as possible to put that data into useful information. Because games are won by a team's players' actions on the field, this information typically takes the form of statistics on players. They use the statistics (information) to evaluate the players to come as close as possible to explaining past ourcomes or predicting future ones.

This in no way devalues the win. In fact, the statheads are saying only, "let's get as much information as possible to evaluate our players." And why evaluate the players? To help determine how teams win games.

The Cy Young is given to the best pitcher in each league each season. All of the information available points to one clear conclusion: there are better ways to evaluate a pitcher's performance than his won/loss record. Does this devalue the win? Of course not. The premise is only that you can't tell who pitched the best by comparing pitchers' won/loss records. While a pitcher certainly contributes to the win (or loss), he doesn't completlely control the outcome of the game. The only thing devalued is the use of a pitcher's win total to determine how well he pitched.

Statheads measure things like a player's contribution to a win with advanced statistics like WARP (Wins Above Replacement Player). WARP puts a number value on a player's overall performance, or his contribution to the team's wins. Is it perfect? Of course not. But it's more information instead of less. But more importantly, it's statistically relevant. Let's put it another way; people argue a pitcher, "pitched to the score" meaning he didn't mind giving up more runs or only threw fastballs with a big lead. But the statheads ask if that is, in fact, true, and then set about to prove or disprove it. (It's mostly been disproved, like the idea of clutch hitting.) WARP is a Baseball Prospectus stat. It (and its many variations) are used in almost every discussion on the BP website ranking players against eachother (including hall of fame discussions, etc.). Is that devaluing a win? No. It's emphasizing wins over more traditional statistics.

Take the MVP for example. If you want to know which player was the most valuable to his team, do you want to know how many rbi's he hit or how many wins he contributed? Of course, you would prefer to know how many wins. Using "advanced statistics" then really means, "using more, and more accurate, information." When an old-school commentator (like Kevin Kennedy) says something like such-and-such player was "clutch" what he's really saying is, "I'm ignoring information that is readily available, through the tremendous effort and hard work of others, to assist in my evaluation of a player." If a pitcher had more control over which team actually won a game, it would be the perfect stat. But there are too many other factors involved to measure pitcher performance that are better than simply asking who won. Most fans only care who won, and that's fine. But in giving out an individual award for pitching excellence, "who won" is not enough information.

By the way, baseball already has an award that, "values wins." It's a team award and it's called the division championship.

No comments: