April 16, 2005
One of the most popular topics in our sabermetrics course is player projection. The "simplest" projection method simply considers the previous three years of performance and then adjusts for a player's age. Basic multiple regression models can then be used to predict future performance. For example, see Tangotiger's ZIPS system, a nice method that yields relatively good predictive capability.
But the data can be tweaked to provide better estimates. For example, the PECOTA model promulgated by Baseball Prospectus uses a database of major league players from the last 50 years to identify "comparables" to current players. Based on age, position, career paths, and phenotype, these historical comparables are used to fine-tune the model to better predict the future. For each major league player, you can receive a list of most-comparable players from Baseball Prospectus, ranked by similarity scores.
Not only are these compariables useful, but they're fun. After all, a handful of numbers from a regression model paints a good picture of what we can expect from Carl Everett in 2005. But it's a much clearer picture if I say, "Think Mike Devereaux, 1997."
Thus armed, we can go around the diamond for the 2005 Sox and look at some of their closest comparables.
Joining Jurassic Carl in the outfield is Aaron Rowand, who has the close comparable of Ellis "Crash" Burks, circa 1992. Burks received a lot of cheers as a crucial cog on the 1993 Western Division Champions, but his league average on-base and slugging percentages were hardly worth clapping over. Mr. Burks, welcome back to Chicago.
Going into 2005, newcomer Scott Podsednik also finds a former Sox as one of his closest comparables – Rudy Law. But this isn't the Rudy Law of 77 stolen bases on a 99-win 1983 team; no, this is Rudy Law from 1986, playing out the final season of his career in Kansas City with 14 stolen bases and a .327 OBP. Optimistic Sox fans hope Podsednik to be many things this year – the White Rudy Law is not one of them.
The fourth outfielder in the mix is another new guy, Jermaine Dye. Though he wears #23, it's not Michael Jordan that he'll remind you of. No, Dye is basically the Polish Prince himself, Rickie Zisk, vintage 1980. And Dye's back-up? Timo Perez (*cough*DaveMartinez*cough*).
Up the middle on the infield, we have Shawon Dunston, 1989 edition, masquerading as Juan Uribe, and his double-play partner Paco Iguchi, just a run-of-the mill secondbaseman of the ilk of Norberto Martin (1997) or Tony Graffanino (2003).
Paul Konerko is at first-base, and while clearly the best player on the team, if you look close enough you might see shades of Greg Walker before his 1989 decline. Better signs for Paulie point toward Beltin' Bill Melton, 1974, or even Frank Thomas (ver. 1.0), 1958.
Speaking of Thomas, who does PECOTA show as the most comparable players up to this point in Big Frank's career? Sadly, Ted Williams no longer makes the list of top 20 comparables. Now it's Jose Canseco, 2001.
Captaining this squad is one of A.J. Pierzynski's closest comparables, Jorge Fabregas, who – as a teammate of Ozzie Guillen – caught 100 unremarkable games for the 1997 White Sox. And speaking of unremarkable, we'll round out the 2005 Sox with some ghosts likely to get some playing time: Lance Johnson 1990 (now, Willie Harris), Greg Norton 1999 (Joe Borchard), and Joel Skinner 1989 (Ben Davis). Too bad Wilson Valdez coudn't crack the club. A Jerry Dybzinski retread would have been fun.
So there you have it, your 2005 Chicago White Sox:
Law, CF
Dunston, SS
Burks, RF
Canseco, DH
Walker, 1B
Devereaux, LF
Martin, 2B
Fabregas, C
And batting 9th, at Third Base, Joe Crede.
Seriously, you could build a team around him. Just don't expect to win 90 games with it.