THE BOOK--Playing The Percentages In Baseball

Wednesday, August 23, 2006

Selective Sampling - How NOT to Choose Players

Cy Morong takes a look at establishing the replacement level.  He says:

I ranked all NL pitchers from worst to best in RSAA per IP

And therein lies the problem.  You cannot establish, after the fact, who is a replacement-level pitcher.  Say you have a pitcher that gets thrown into the dustbin, I dunno, maybe Orlando Hernandez, between 2003/2004.  He was a free agent, and signed with the Yanks for 500,000$.  That’s the definition of replacement level.  In 2004, he went 8-2, with a 3.30 ERA.  However, he would fall off Cy’s (and pretty much everyone else who looks at this issue) selection criteria.

This is a huge selection bias, and invalidates any results presented.