Groundballs and the Overall Picture

I have seen people claim that a pitcher’s ground ball rate is not a useful piece of information because grouping pitchers by it shows no meaningful difference in runs allowed. FIP is comprised of a pitcher’s strikeouts, free passes (walks plus hit batsmen) and home runs. Based on the last few posts the data seems to indicate that strikeouts and free passes are not meaningfully effected by a pitcher’s ground ball rate and home runs decrease. A scatter plot bears out the expected result.

Instead of artificially grouping pitchers, a full trend line points to a pitcher’s ground ball rate being a useful piece of information, even on its own. It is not just FIP though but actual runs scored follows the same trend.

In conclusion from the previous five pieces here is a a list of variables that appear to have no meaningful deviation as a pitcher’s ground ball rate increases:

  • Home run rate per non-groundball
  • Overall strikeout rate
  • Overall walk rate
  • Slugging percentage on line drives in play.
  • And here is a run down of what I consider the be the key results found for what does happen as a pitcher’s ground ball rate increases:

  • Slugging percentage on fly balls in play increases.
  • Overall slugging percentage on non-ground balls increases.
  • Less pop ups are allowed.
  • More runners reach via error.
  • Fewer home runs are hit.
  • Their FIP goes down.
  • Their RA goes down.

  • Print This Post

    Matthew Carruth is a software engineer who has been fascinated with baseball statistics since age five. When not dissecting baseball, he is watching hockey or playing soccer.

    24 Responses to “Groundballs and the Overall Picture”

    You can follow any responses to this entry through the RSS 2.0 feed.
    1. William says:

      This is an awesome series. As a National fan (really), I have been worried about their GB pitcher/poor infield defense ratio, and only now wonder if you might look into the importance of said-D on making these types of pitchers so favourable.

      At any rate, great stuff, thank you…

      Vote -1 Vote +1

    2. Bob Loblaw says:

      There is one other factor I wonder about — pitch efficiency. Do GB pitchers, by pitching to contact, throw fewer pitches per batter faced, and thereby go deeper into games?

      Vote -1 Vote +1

    3. Greg says:

      I’m no statistician, but isn’t your R value too low to show a correlation between FIP or RA and GB%?

      Vote -1 Vote +1

    4. Eric Feczko says:

      They are R^2 values, which describe the amount of variance explained by the correlation. A correlation of .07 means that 7 percent of the variance in FIP is explained by the groundball rate. So yes, these values are too low to make the statement that FIP (or runs allowed) are related to groundball rates.

      Given the number of datapoints, it is possible that the result is statistically significant, but I can’t see how the result is meaningful in any way.

      Vote -1 Vote +1

      • notdissertating says:

        Low R-squared values do not imply there is “no relationship.” The correct inference is that ground ball rates of pitchers do not have a lot of predictive power for FIP. This statement is entirely consistent with the statement that on average high ground ball rate pitchers have lover FIPs. The distinction is between means and variances.

        Vote -1 Vote +1

    5. Brilliant work Matthew. What program do you use for these charts? They don’t look like excel graphs.

      I am bookmarking this piece for the next entry into QSE

      Vote -1 Vote +1

    6. Michael says:

      The correlation on these graphs is horrible. If I got a graph like this in my physics experiments, I would be forced to say that there is little to no correlation to what I was graphing. I think the conclusion should be that groundball rate has little to no effect on FIP or RA.

      Vote -1 Vote +1

    7. KJ says:

      I have to agree with both Michael, Notdissertating and Eric. It is difficult to make an inference here on the relationship between ground ball rate and FIP. Matthew, what was the r coefficient (correlation coefficient)? I’m mostly curious. I would ask that readers not make any assumptions based on this data.

      Vote -1 Vote +1

      • Eric Feczko says:

        Thanks, Notdissertating, for clarifying my post.
        The r coefficient should be the square root of the r^2 value. I’m assuming a pearson’s coefficient was used (as the data look normally distributed).

        Notdissertating, although this doesn’t address the question of “on average, do high groundball rate pitchers have lower FIP?”, that question has been asked and answered via a comparison of means previously (as Matt indicated at the top of his post); no significant differences between “high” GB and “low” GB pitchers was found (although “high” and “low” were arbitrary terms). To be fair, I don’t think that analysis (as matt points out) was performed properly either. The way to check would be to plot the distribution of GB rates, pick the tails of the distribution to form the “high” and “low” GB pitchers, and then perform a one-tailed t-test on FIP for the two groups.

        Vote -1 Vote +1

    8. Daern says:

      How are you making these graphs?

      Vote -1 Vote +1

    9. KJ says:

      the r value should be .27 I was wondering if that is what Matthew got. That would mean that there is a correlation about 27 percent of the time. Note:That statement is a crude one and not exactly accurate but it gets my point across. Some of the others could probably do a better job of explaining. Anyway, a .27 r value indicates a weak relationship.

      Vote -1 Vote +1

    10. vivaelpujols says:

      C’mon guys, this isn’t a controlled physics experiment. It’s an uncontrolled real world experiment trying to pick up one effect among at least 10 significant others. There is no “threshold” for what a good R2 should be. It’s ridiculous to categorically say that a “.27 r value indicates a weak relationship”, especially when those speculating can’t even see the significance or standard errors around the slope.

      Vote -1 Vote +1

    11. tangotiger says:

      Matthew: can you show FIP minus ERA, relative to GB rate?

      If I am correct, then we should see a fairly flat line.

      Vote -1 Vote +1

      • I would, but I don’t have a database set up yet that keeps track of ERA. But inferring based off the FIP-RA graph and knowing that as GB% goes up, unearned runs also goes up, I would have confidence that you are correct and the slope would be fairly flat

        Vote -1 Vote +1

        • Greg Foley says:

          Hi Matthew. There seems to be a general consensus among us commenters that the R^2 values are too low for the correlation between FIP and GB% to be meaningful. If we’re all wrong for some reason or another, could you please enlighten us? Thanks.

          Vote -1 Vote +1

    12. CircleChange11 says:

      Another thing to understand about “groundball pitchers” and “pitching to contact” … They are generally pitchers with “lesser stuff”. If guys *could* be strikeout pitchers, they would.

      So, where a groundball pitcher may pale in comparison to a top 1-2 starter on each staff … by “pitching for groundballs” he may be significantly better than he would be if he pitched “like everyone else. This is where we would like to be able to compare Pitcher A with one approach versus Pitcher A with a different approach, rather than comparing individual pitchers of various types and quality with each other.

      Note: By ‘pitching for groundballs’ I am speaking primarily of pounding the strikezone with late breaking/sinking/tailing pitches.

      “Pitching to contact” isn’t so much a philosophy choice, but rather the result of a pitcher’s stuff, and the acceptance of what/who he is and using his quality to his advantage. I have coached pitchers whose stuff is so good that they literally cannot “pitch to contact”, due to their combination of velocity and movement. We see similar guys in MLB (relative to their level). It kills me to hear comments suggesting that high strikeout guys should “pitch more to contact” as if the hitters weren’t trying to make contact. *Shrugs*

      The KEY is that by focusing on the GB, the GB Pitcher finds a way for him to be successful because his stuff isn’t likely good enough to pitch any other way and survive.

      There are some very interesting things in this study. I figured GB pitchers would give up less walks, more hits, and fewer pop-ups. HR’s allowed, i wasn’t sure about. They could be equally susceptible on mistakes, whereas a power pitcher often gets away with “pitching up” b/c their stuff is overwhelming.

      Again, we have to remember we’re talking about a group of pitchers whose “stuff rating” is probably around a 6.5 to 7.5, and they (as a group) are rather successful in a league dominated by pitcher’s whose stuff rating would be more in line with 8.0+.

      VERY interesting stuff in this groundball series. Thanks for the discussion.

      Vote -1 Vote +1

    13. Jim says:

      Are the higher slugging percentage results on flyballs caused by or part of the groundball/flyball platoon differential (e.g. flyball hitters generally hit groundball pitchers better than flyball pitchers)?

      Vote -1 Vote +1

    14. Barry Reed says:

      I’m probably missing something, but looking at the previous post on walks, it seems to me that BG rate has a similar impact on walks and FIP… For whatever reason, in the walk graph, the equation is based on x in integers, while the FIP graph’s equation uses x in decimals (i.e., 25% results in x = 25 for walks but x=0.25 for FIP)… If you put them on the same scale, the slope for walks is 3.72 and for FIP it’s 2.98… The R^2 values are all low, so what am I missing?

      Vote -1 Vote +1

    15. Pompous but earnest grammar guy says:

      You mean “affected” not “effected”.

      Vote -1 Vote +1

    16. Barry says:

      You’ve got to love this mate!

      Vote -1 Vote +1

    Leave a Reply

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

    You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

    Current day month ye@r *