Proceeds from the ads below will be donated to the Bob Wuesthoff scholarship fund.

Monday, November 22, 2004

Don't Mean Beans To Pinto: More On "Contactball"

Continuing the debate on "Contactball", I noticed David Pinto had a followup on his blog, looking at Steve Lombardi's discussion. There are several important points here:
  1. Given two teams with identical OBP, the one with the higher batting average is going to score more runs. This was a point Pinto made earlier.
  2. Lombardi's FSC stat isn't what he thinks it is. What he wants to measure is
    FSC2 = -----------
    but instead he has
    H * ((AB+BB+HBP+SH+SF)-SO-BB)
    AB * (AB+BB+HBP+SH+SF)
    The key is that he uses TPA as the denominator for his RC/G analysis of frequently-walking teams, but fails to do the same with high batting average teams. The second formula leaves him with an AB2 term in the denominator, but this glosses over an even more important question: why do you care about the percentage of hits times the percentage of contact?
  3. He doesn't actually do correlation analysis on the results. The point of the game is to score runs, period. The only way to know if a certain strategy works is to see whether it results in more runs scored. The first -- but far from necessary -- test is correlation. If you can't see correlation, you won't be able to prove causality. Lombardi just looks at a few select teams and observes that "teams who were good at frequent contact while hitting safely often are better at batting than teams who used the walking often approach". Taking a look at correlation for the two even limited to the 11-year period 1994 through 2004 gives us these results:
    Stat    r
    OBP   .684
    AVG   .596
    FSC   .452
    FSC2  .439
    Once again, the connection to runs scored fails to materialize.
I understand what he's trying to do; I just don't see that he's shown what he thinks he has.

Is there any correlation to runs scored with injuries (i.e. do injuries have a large negative effect on runs scored) and is there any correlation of runs to % of plate appearances by the top nine batters (or 8 in the NL) (i.e. does a team with a set line up score more than a team that always interchanges players due to injury or performance). I know the Red Sox of '03 and the Angels of '02 in particular had a very large percentage of at bats for their top guys and scored a lot. I am just wondering if there is a general correlation there.
Interesting. I suppose you could come up with a proxy for that (divide players into high-OBP guys and not, and then look for guys with lots of ABs), but the general trends would agree with your hypothesis.
I think even if there is a correlation (seems likely) the question then becomes: is it due to young guys used as fill-ins don't walk as much (fitting in with the OBP vs. AVG correlation), is it in lineups and players, is it team depth, or is it some combination of these and other things. I don't know how you might do a quick and easy test to differentiate these things.

Post a Comment

Newer›  ‹Older
This page is powered by Blogger. Isn't yours?

WWW 6-4-2