Pinch-Hitting Blogger: Ultracyberstatman

Once again, we're digging deep into the shitpile of nobodies that is the World Wide Web to lift another Yankee blogger from obscurity.

Rex Hibbard-Staley, a statistician extraordinare from Elmira, NY, and captain of the Glider City KFC Bowling Bucketeers, says he is "not only responsible for serving upstate New York's tastiest chicken breast-crunching, but also its most succulent and sexy number-crunching, too!" We asked Rex to get out his calculator and give us the lowdown on the Melky trade. He filed this report:

The basic idea of hierarchical batting average modeling (also known as multilevel BAT modeling) is to elevate lowest-level Yankee units into a hierarchy of successively higher-level Yankee units, or Yanknits, or in this case, Melknits, for simplicity.

For example, hypothetically, Melky plays CF, in the OF, on NY, in USA. By co-signing individual Melky at-bat performance outcome Melknits on an empiracal Bayes curve -- (not to be mistaken for a Josh Beckett curve, ha ha!) -- as a sum of effects for individual Melknit outfield output, for her/his team, for CF, OF, NY, et al, we get a value of 3.04.

I also applied Melky regression coefficients, which were specified inferences easily drawn from available data for the Melknit population means at any level (CF, OF, NY, USA, NA - that's North America, etc.) from a frequentist perspective as "Best Linear Unbiased Predictors" (BLUPs: see Robinson, Statistical Science, 1991, pp. 15-32). Raw Melkian properties ran consistently 0.03 higher than simple sample-based estimators from the Brett Gardnit model (3.01), making Melknits a useful player tool (see Ghosh and Rao, Statistical Science, 1994, pp. 55-76).

This sequencing suggests for the Yankees to have achieved a 3.07-3.09 rating, guaranteed World Series, they should have kept Melky and signed John Lackey.