GROCHOWSKI: Great time of year for MLB statistics junkies
BY JOHN GROCHOWSKI For Sun-Times Media
Colorado Rockies' Troy Tulowitzki, left, leaps over San Francisco Giants' Brandon Belt (9) after forcing Belt out at second base during an MLB spring training baseball game, Friday, March 22, 2013, in Scottsdale, Ariz. (AP Photo/Ross D. Franklin)
Throughout the early
portion of the baseball season, two books I keep near as I watch the games are the annual editions of Baseball
Prospectus ($24.95, Wiley and Sons) and Baseball Info
Solutions’ Bill James Baseball Handbook ($26.95, Acta Sports). The James book is light on text but crammed with information of the numerical kind, while the Baseball Prospectus tome includes blurbs about teams and players along with the stats.
One thing they have in common is a statistical look into the future. Baseball Prospectus uses the PECOTA system developed by Nate Silver — he of election-analysis fame — to project what each player’s 2013 season will look like. The Bill James Baseball Handbook uses James’ system for hitters, while much of the heavy lifting for pitching projections was done by John Dewan.
No one claims 100 percent accuracy; that’s not possible where human beings are concerned. Where there is large error, playing time usually is a factor (somebody earns an every-day job, someone else has production drop enough that he loses his spot in the lineup). But there are broad arcs to player careers that point in the direction of probabilities. Teams that use sabermetrics in personnel decisions understand the arcs.
Production tends to improve through about age 27 or 28, and nearly all players are showing some decline by age 30. Home runs and walks tend to increase for a few years beyond a player’s overall peak. The best predictors of how a player will do this season are his most recent seasons.
There’s more, but you get the idea. No one’s ride is perfectly smooth. There are ups, downs and detours. But players’ pasts point in the general direction of their futures.
Part of the PECOTA formula is the use of similarity scores, looking at the most comparable players in baseball history. Given a player’s three most recent seasons, how did the most similar players in baseball history perform?
What does the future hold for key White Sox and Cubs players? Without going into the detail you’ll find in the books, let’s look at a few:
† Starlin Castro, .289 batting average, .329 on-base percentage, .427 slugging percentage by PECOTA; .304/.346/.448 by James. Each system sees Castro with 12 homers. His most comparable players listed in Baseball Prospectus are Jose Reyes, Barry Larkin and Wil Cordero.
† Anthony Rizzo, .257/.324/.470 by PECOTA; .283/.346/.517 by James. PECOTA sees 17 homers, James 33. That’s largely explained by playing time. PECOTA is based on about the same time Rizzo had in his partial season in 2012, while James projects 604 at-bats.
† Adam Dunn, .222/.348/.457 with 30 homers, 90 walks, 182 strrikeouts by PECOTA; .207/.341/.442 with 33 homers, 103 walks, 213 strikeouts by James. PECOTA projects a small decline in playing time, from 649 plate appearances to 579.
† Paul Konerko, .277/.361/.484 with 26 homers by PECOTA; .281/.367/.486 with 30 homers by James. Both projections see similar power but small declines in batting average and OBP from Konerko’s .298/.371/.486, 26 homers of last season. At age 37, that’s only normal.