MARCEL the Matrix
One of the best known player forecasting systems is Tom Tango’s MARCEL. One of MARCEL’s primary benefits is that it is relatively simple to understand. The basic idea is to take a time-weighted three-year average of past performance and adjust for age. This simple projection system has proven to be relatively robust, in that many others have tried and failed to best MARCEL’s projections. Today, we’ll calculate MARCEL in R.
Being a relatively simple projection system, MARCEL does not require play-by-play data. Thus, we can compute in R using the venerable Lahman Database, which is packaged into R by Michael Friendly. The Lahman Database contains seasonal totals for all major league players going all the way back to 1871.
Part of this exercise will involve confirming that we have done the calculations correctly by comparing our results to Tango’s. In his original instructions, Tango computes MARCEL for…
View original post 1,092 more words