[Previous entry: "Pictures 17: Bridging Hell Gate"] [Next entry: "Pictures 18: Not Buying It"]
12/17/2004: "Baseball Projections 2: Flaws"
That seemed like a good idea.
I think I was blinded by the fact that I got what seemed like reasonable numbers for Pedro Martinez. Working with my formula, though, I came up with the following projections for a pair of pitchers who would seem likely to accomplish much more than these numbers:
Randy Johnson: 11-9, 153 IP, 60 ER, 43 BB, 130 K, 3.52 ERA
Mike Mussina: 10-8, 146 IP, 63 ER, 39 BB, 93 K, 3.86 ERA
For Johnson, it seems unlikely that he'd replicate Luis Tiant's 2-2, 29.2 IP, 5.76 ERA performance with the 1982 Angels or Orel Hershiser's 1-5, 24.2 IP, 13.14 ERA with the 2000 Dodgers. It's probably unwise to factor in Lon Warneke's 0-1, 14 IP, 3.86 ERA season for the 1945 Cubs or Jim Bunning's 4-14, 160 IP, 3.88 ERA effort for the 1968 Pirates when trying to figure out what Mussina will do.
It's certainly worthwhile to try to derive projections from comparable players: that's a key element of the PECOTA projections that Baseball Prospectus runs. Unfortunately, I don't have the appropriate software to tinker with my own formula and generate sensible projections of my own.
I thought it would be a really cool thing to be able to come up with projected 2005 numbers for players as they moved from team to team throughout the winter. It would be something to add to the discussion of the hot stove league. I just don't have the tools at my disposal right now to do it in a way that makes sense or produces good numbers -- even if my first go at it made me think I did.
Back to the drawing board.
About
