My writings about baseball, with a strong statistical & machine learning slant.

Monday, April 5, 2010

Team Pitching Projections (IV: San Francisco Giants)

Unlike the Rays, my system and CHONE disagree strongly on our expectations for the SF Giants. Individual projects can be found here.

Last year the Giants compiled 264.4 pitcher VORP, good for 3rd in MLB. This propelled them to an unlikely Wild Card run, despite having no hitting beyond Pablo Sandoval.

The one Giants game I attended last year must have been a typical win for last year's squad. Tim Lincecum threw eight shutout innings, striking out eight, walking one, and giving up only seven singles. The Giants' hitters looked just as hapless against Joe Blanton as the Phillies looked against Lincecum. They won 2-0 on two Juan Uribe sac flies. Brian Wilson pitched the ninth for his 27th save, and the Giants' secondary pitchers never even warmed up.

Neither my system nor CHONE expects the Giants to repeat their amazing pitching performance this year. CHONE projects them to fall to 134 VORP. That would be the same value as last year's Royals staff (substitute Lincecum for Greinke, Brian Wilson for Soria, and Barry Zito for Gil Meche). My system, however, regresses the Giants' projection down to 189.7 VORP only, which would still make them about 20-30 runs better than the MLB average.

The reason for this 55 run difference is that I'm more bullish on each of the Giants' top four starters:

  1. Tim Lincecum: 225 IP, 3.02 ERA (196 IP, 3.12 ERA for CHONE)
  2. Matt Cain: 197 IP, 3.89 ERA (188 IP, 3.93 ERA for CHONE)
  3. Barry Zito: 148 IP, 4.32 ERA (169 IP, 4.85 ERA for CHONE)
  4. Jonathan Sanchez: 163 IP, 3.92 ERA (151 IP, 4.47 ERA for CHONE) 

I was going to write about how Jonathan Sanchez has been much better than his raw ERA and losing win-loss records show, and that at 27, we should expect him to be at least as good as he was in 2008-2009, but probably much better. However, Peter Bendix of RotoGraphs explains all of this much better!

I am similarly puzzled by CHONE's low projection for Barry Zito's ERA. Despite the really bad contract, he's still a useful starting pitcher, and has consistently had a ERA below 5.0, except for his poor 2008 season. At 31 he isn't that old yet, and has thrown 180IP+ each of the past 10 years. If anything, my 148 IP estimate looks low, while the 4.32 ERA looks about right.

Enough has been written about how Matt Cain's 2.89 ERA last year was due to unsustainable good luck with BABIP and with stranded runner rates. Both me and CHONE agree that he should be good, but not elite, in 2010.

Tim Lincecum was easily my highest VORP projection for 2010, and also one of my lowest ERA projections, even with relievers included. As Tom Tango pointed out in his blog, projecting a 3.04 ERA means that there the same chance that he will post a 2.54 ERA as the chance that he will post a 3.54 ERA. If so, that would make today's Tim Lincecum a value similar to mid 1990's Greg Maddux! Does that seem plausible? My projection system seems to think so.

1 comment:

  1. Hello Ivan!

    I am a huge Giants fan and stumbled upon this article while reading about our championship years :)

    I am a first year at UCSC studying Computer Science. I want to get involved in Machine Learning, especially revolving around baseball.

    I was wondering if you had any advice/tips for me to get exposure in this interesting field?


    Ayush Arora