In a field of science as imprecise as measuring the performance of a baseball player I think 17% is probably acceptable.
The underlying question is, can the margin of error be made any lower? If yes, I'm sure it will be done.
If no, we all just have to accept it for what it is.