Thursday, July 1, 2010

How accurately can we estimate a hitter’s runs?

The short answer, according to the Hardball Times: Pretty accurately, but not perfectly. I wanted to link to this article by Colin Wyers because I thought it was important. Yet I didn't see much discussion about it when it came out and I hang out online in some of the some places that statheads do. Here's the important part:

Some things to take away from the chart:

Power hitters are far more "inconsistent" than other kinds of hitters. Doubles, triples and home runs by far have the highest standard error.
Walks are very consistent, showing very little standard error compared to other events. Outs are also very consistent. So, all else being equal, players with either very high or very low on-base percentages would tend to have less error in their linear weights estimates.

Colin was writing about hitting. That's the portion of baseball that's easiest to analyze statistically. If there's fuzziness there, what about when it comes to analyzing the rest of the game? None of this should suggest that Duane Kuiper was a better hitter than Barry Bonds or anything like that. But I am curious to hear what others like Tom Tango have to say about this article. Maybe I missed the discussion.


  1. There are three major reasons why his results didn't rock any boats:

    1. We have the data now to look at the change in run expectancy from the PBP data (Fangraphs publishes it as RE24). Of course, the RE itself is subject to error.

    2. There is still a point of view that the actual situations the player is put into are out of his control, and so the best estimate of his true talent is the estimate based on standardized event values.

    3. Over the long haul, unless players demonstrate different ability in different situations, the errors should even out. And consistency is not necessarily a good or bad thing--it depends. If low OBA hitters are more likely to match their RE24 to their linear weights, that means they are consistently bad (at least as a group).

  2. Those are valid criticisms, but I didn't see anyone raise them in August. As for point two, I'm in the same boat as (I think) Pat Andriola.* I'm more interested in what happened in the past than predicting the future.

    * There was an article in my Google Reader recently about using FIP vs ERA (or something similar.) I honestly forget who wrote it, but I think it might've been Pat.

  3. I don't really consider them criticisms--I don't disagree with Colin's conclusions or methodology. And I too do not recall any discussion on the article.

  4. I guess the word I was looking for was critique. ANyways, thanks for the posts and I enjoyed yours about the LL Bean scorecard.