Friday, March 31, 2006

Great American Ballpark is Not a Hitter's Paradise

As I mentioned in last night's post, I have really enjoyed Orel Hershiser as a commentator over the past two BoSox ST games I've watched. He is far more insightful that most of the color guys they have in that booth and brings a knowledgeable coach's perspective. But there was one thing he kept harping on that I found quite frustrating: his argument that Great American Ballpark is a severe hitter's park. He repeatedly compared it to the Ballpark in Arlington, which he cited as a major reason why the Rangers have had trouble developing a solid pitching staff over the years.

He may be right about Arlington, but the data don't support his argument for GABP. Let's do a quick comparison of these two parks. I'll also throw in Coors Field, as that's the other park that was mentioned in that conversation (thank you Marty).

First, let's look at the composite park factors reported in Baseball Prospectus's 2006 Annual, along with their "categorization" on that park. These park factors can be read as the adjustment one should make to a team's offense when they play in this park. A value below 1.0 indicates that a team will score fewer runs that normal, while a value above 1.0 indicates that teams will score more runs than normal.

GABP: 0.988 (slight pitcher's park)
Arlington: 1.067 (severe hitter's park)
Coors Field: 1.122 (severe hitter's park)

Which one is not like the others? Coors Field has a longstanding reputation as a ridiculous place to play baseball because the offense is so inflated. Arlington, similarly, has a well-justified reputation as a hitter's park. But what's going on with GABP? All we ever hear about it is that it is highly conducive to homers, which is supposed to be why Eric Milton has gotten killed there (which is partly, though probably not entirely true). Why isn't that showing up in the data?

The Hardball Times' 2006 Annual has an excellent article by Dave Studenmund about ballparks and how they affect the game. In it, he cites a superb set of park factors that are available on the web by a guy called U. S. Patriot. If you go to his site, you'll find the park factors below. These are based on multiple years of data for each park, corrected for sampling error (regressed to the mean), and are separated into a park factor for total runs allowed (RunPF) and a park factor for home runs (HRPF):
Park        RunPF   HRPF
GABP: 99 106
Arlington: 106 107
Coors Field: 115 114
In Patriot's data, 100 is average. Therefore, GABP is still showing a slight negative effect on overall runs scored. Nevertheless, here we do see that home runs are positively influenced by GABP...but not as badly as in Arlington or Coors. Since HR's are inflated, other sources of offense must be concurrently depressed in GABP in order to result in a net neutral park in terms of overall runs scored.

To pursue this further, I'll refer back to Studenmund's Hardball Times chapter where he reports some novel data on batted ball types and results in different stadiums. He reports the average value, in terms of the number of runs, of three batted ball types in each stadium: outfield fly balls (OF), ground balls (GB), and line drives (LD). A value above 0.00 indicates that this batted ball type results in more runs than in a typical stadium, while a negative value indicates that this ball type results in fewer runs than average. Data on our three stadiums:
Park         OF      GB      LD
GABP 0.016 -0.021 0.005
Arlington 0.057 -0.004 0.017
Coors Field 0.077 0.003 0.016
Here we see that GABP is very rough on ground balls, with each one resulting in 0.021 fewer runs than average. This was, in fact, the most negative ground ball effect among all ballparks in the article! In contrast, the average ball hit to the outfield, as well as the average line drive, results in more runs than average. Nevertheless, these two effects were far smaller than in Arlington and Coors. In fact, the following ballparks had larger positive run effects on fly balls to the outfield (OF) than GABP: Coors (rockies; 0.077), Arlington (rangers; 0.057), Wrigley (cubs; 0.045), Fenway (bosox; 0.043), U.S. Cellular (white sox; 0.040), Rogers Centre (blue jays; 0.026), and Minute Maid Park (astros; 0.019). Yes, most of these have reputations as hitters parks, but the key point here is that we're not dealing with something completely out of this world in GABP.

Conclusion & Looking Forward

Yes, GABP is permissive when it comes to home runs. But it's not uncontrollable. The problem we're encountering has more to do with the pitchers we have on our team than anything else. As these data indicate, pitchers that induce fly balls more often than ground balls (last year = Milton) will struggle in GABP, just as they would at Wrigley, Fenway, and U.S. Cellular.

But as each of the teams at those three stadiums have shown, it is very possible to build effective pitching staffs in those ballparks. The key is to focus on pitchers that can exploit the ballpark's strengths. Great American Ballpark is, in Studenmund's words, "death to ground balls," with the most negative effect on ground ball-based runs of any park in baseball. If we build a team around pitchers who can keep the ball on the ground, we can exploit this characteristic of our park in our favor. Last year, Harang, Arroyo, Claussen, and Williams all allowed more ground balls than fly balls to the outfield (ok, Arroyo allowed equal percentages of each; source=hardball times '06 annual, pp.330-332), which gives me some hope for our rotation moving forward. -j

8 comments:

  1. That's really interesting. But I have one sort of speculative question. What exactly makes one ballpark worse than another when it comes to groundball-based offense? And specifically, what do you think it is about GABP that has this affect. With HRs, it's easy to point to outfield dimensions, wind conditions and altitude (in one case) to explain why some parks get more HRs than others. It's not so clear with groundballs.

    ReplyDelete
  2. Very good question, and I really don't know the answer to this. It could be as simple as the habits of the groundscrew in terms of the height/density/type of grass used, the slope of the infield, etc. I can see the weather having some sort of effect as well...humidity x grass interactions or something. But I don't really know.

    What I can report is that a) those groundball effects were based on 4-years (or, in GABP's case, 3 years) of data. Furthermore, the authors reported reasonably strong (though not overwhelming) year to year correlations for those numbers. So whatever the effect is, it's unlikely to be due to chance alone. Unfortunately, I don't have the data in front of me or I'd report the actual correlation (I'm at work). -JinAZ

    ReplyDelete
  3. I have the same question. If it is death to groundballs, is it the park? Could it be the IF defense? Have the Reds IF's been that good? Have the same players been there for a few years? Or is it really the grounds crew?

    ReplyDelete
  4. Hi Lars. It shouldn't be that sensitive to defense, unless the defense has happened to play differently at home and away. These park factors are calculated by evaluating both how many runs the Reds score at home vs. away as well as how visiting teams play at GABP vs. elsewhere. Therefore, unless there's a big difference (by chance) in how defense has happened to perform at home vs. away, the Reds' defense shouldn't affect the park estimations. That's part of the reason they use multiple years to calculate these park factors. It reduces the effect of chance events.

    Although, as it happens, the Reds defense (both infield and outfield) has been below average for several years now. Aside from Encarnacion, all (I think) Reds infielders who have played a substantial number of games in the infield defense over the past several years have been had below average range ('05 team runs saved = -0.32 --> http://www.baseballmusings.com/archives/cat_probabilistic_model_of_range.php).
    -j

    ReplyDelete
  5. I'm still having a hard time believing that GABP is easier on GBs versus FBs, though you certainly seem to have the stats to prove it. Does this mean we should seek FB pitchers over GB guys? If so, are there FB pitchers that give up relatively few HRs?

    If FB pitchers are the way to go, then Williams and Arroyo were good pick ups. And maybe Milton will rebound. Milton, as bad as he was last year, actually gave up fewer HRs than the previous year in Philly, where he pitched respectably (201 IPs, 4.75 ERA). His BABIP was really high, indicating that maybe bad luck and defense played larger roles in his 2005 floggings than most thought.

    ReplyDelete
  6. Hi Ken,

    I think you may have it backwards: GABP has a high park factor when it comes to balls hit to the outfield, meaning that those balls tend to convert into runs more often than at an average park. HR's are also positively effected by the park.

    In contrast, ground balls tend to result in fewer runs and more outs than at an average park. This means that it is absolutely in the Reds' best interest to seek out ground-ball pitchers. Fly ball guys like Milton and Williams are probably always going to struggle in our park...though as you point out, other factors contributed to Milton's awful year last year--he was just as bad on the road, and you can't blame GABP for that:
    http://jinaz-reds.blogspot.com/2006/03/on-eric-milton.html).
    -j

    ReplyDelete
  7. Gotcha. I'll have to temper my optimism for Arroyo accordingly, though the two spot he just gave up to the Cubs also did that.

    ReplyDelete
  8. Ahh, I thought I still had an hour before the game started! ::fumbles for cbs gamecenter::

    ReplyDelete