Tuesday, April 12, 2005

Solving the Fielding Problem

No, not the Nats' fielding -- so far that has been pretty good. I'm talking about the problem of evaluating fielding through statistics. The fundamental issue is one of subjectivity. No matter how you cut it, assessing fielding requires a judgment about whether the fielder "should have" made the play, according to a subjective standard. I recall reading an article from the late 1980s where several statheads concluded that we should get rid of the Error stat completely, because it was administered so subjectively around the league as to provide no meaningful information.

ERV scoring allows you to assign an objective value to any event in the game, by comparing the state of the game (the ERV) before and after the event. You can follow the official scorer and assign a value to errors. I would like to assign values to both errors AND good fielding plays, like throwing a runner out at home or robbing a homerun or diving to stop a grounder from going through for a hit. But both Dexys and SuperNoVa rightly expressed concern about my judgment in determining which plays were "good."

What to do? Here's my idea. Turn the task over to the Wisdom of Crowds -- have as many people as possible score the game and make independent decisions about the whether the play is good or not. Just as the value of a stock (an inherently subjective task) is determined by the judgment of independent buyer/seller transactions, the most accurate assessment of fielding will be made by the collective wisdom. ERV scoring provides a good common language for scorers to use in expressing the value of the plays, which can be compared to others. You could average the ERV values or go democractic and adopt the most prevalent view, but in either case you will have a good assessment of the fielding event (and the baserunning event, and other events that require subjective judgement about their value).

Yuda asks here whether the ERV figures, such as Friday Morning Figures, reflect our judgments or those of the official scorer. SuperNoVa follows the official scorer. In contrast, I have chosen to try to reflect good plays, and have deviated from the official scorer on occasion. So the Friday Morning Figures reflect that judgment. As a small first step towards implementing my idea, what I will do each Friday is identify those events where I made such judgments, and you can tell me where I'm wrong (and others I missed), and I will correct where it appears the crowd sees it differently.


Post a Comment

<< Home