March 7, 2011

Four Inning Pitchers

I interviewed Greg Rubin at the MIT-Sloan Sports Analytics Conference. His poster presented research into using pairs of starting pitchers, each expected to go four innings in a start.

2 thoughts on “Four Inning Pitchers

  1. Devon & His 1982 Topps blog

    Wow, this would’ve saved a team 96 runs in 2009? That’s a big swing in team W-L when you look at it in a pythagorean view. This could be huge for a lot of middle of the pack teams that just aren’t making it over the hump… like say, the Tigers, Blue Jays, Braves, or Marlins. It would also save them tons of $$$, being they wouldn’t have to pay for big $$ starters. This could, for a little while, bring small market teams right back into the top of their divisions.

    ReplyReply
  2. Dan

    This is a very interesting idea. I wonder how the results would be in a larger sample size. If I understand him correctly, he calculated this based on the 2010 performance of 8 pitchers. However, just looking at the splits of a few pitchers, I discovered that their performance by inning varies widely. Some actually do better in later innings than earlier. Especially, a number of pitchers have poor stats in the first inning (perhaps skewed by terrible games in which they were pulled early). Would a four-inning workload really result in getting better value out of most or all pitchers–even most or all mediocre pitchers?

    This raises the question for me: are there better ways to tell when to pull a pitcher from the game? Right now a pitcher is generally pulled either when he starts to give up runs or based on a pitch count. Rubin suggests a set inning. But all three of these are rather crude measures. Is there a way to predict one should remove a pitcher from the game, before he actually fails, based on deeper analysis of his performance so far?

    ReplyReply

Leave a Reply

Your email address will not be published. Required fields are marked *