Groundballs and Slugging Rates
Beyond noting that ground balls are less damaging on their own than fly balls I also looked into the effects that a pitcher’s ground ball rate has on his home run rates and found no compelling evidence to support the theory that ground ball pitchers suffer higher rates of home runs on the fly balls they do allow. There’s more to damaging run scoring than just home runs though. I went digging further and started looking at slugging percentage allowed on the batted ball types, absent home runs.
Looking at line drives that ended up in play, were they more dangerous as a pitcher’s ground ball rate increased?
Nope. As it turns out based on the data that I have looked at is that as a pitcher’s ground ball rate goes up, the quality of line drives that he does allow, both the ones that go for home runs and the ones that stay in the park, do not change. Moving on to fly balls however, the first negative trend for ground ball pitchers emerges.
Interestingly, though the rate of fly balls that go for home runs appears to fall as a pitcher’s ground ball rate increases, the slugging percentage on the fly balls that stay in the park goes up. Why? I don’t know. Maybe outfielders shade in and more fly balls go over their heads. Maybe there is some systemic shift in the type of fly balls allowed, though it would be weird for them to be hit harder and yet not see an increase in home runs.
All told, it does look like ground ball pitchers see a rise in their slugging percentage allowed on non-ground batted balls. The effect is neither large nor overly consistent but it appears to be present. Up next, I look at strikeouts and walks.
1
My first guess is that Outfielders are cuaght on their heels more often with ground ball pitchers.
At least some, and probably a great deal of the lower SLG% on fly balls for low GB rate pitchers is a selection effect. Among pitchers who give up a lot of fly balls only those pitchers who produce easy to catch fly balls make it to the majors. This is a necessary skill for a fly ball pitcher. For a ground ball pitcher what happens to fly balls matters much less and so they don’t get weeded out of that basis.
That said, I am also surprised that ground ball pitchers do not have higher HR on fly ball rates. My selection story would definitely predict that there should be.
Neat stuff.
I also am surprised that GB pitchers do not give up a higher HR rate, as some of the extreme GB pitchers I have looked at seemed to.
One of the most important things that needs to be done in the whole area of working with FB data is breaking out IF FB data into its own category since these are virtually always outs and this may well be a skill – whether it is an individual skill or a skill that can be assigned to a type of pitcher or hitter needs to be determined, but serious analysis cannot continue to simply group them together.
Matthew used GameDay data, not FanGraphs, for this series of analysis. In GameDay, FB and IFFB’s are counted separately, and I would be surprised if Matthew included IFFB’s in these studies.
Kind of a sloppy article grammar-wise. I would at least expect the articles to be proof-read before being posted. ‘Then’ instead of ‘than”? Really?
Please. These guys write because they love baseball and they love knowing everything they can about it. If you want to point out a grammatical mistake, just do it – you don’t need to be condescending.
Grammar stuff usually bothers me too. But it’s impossible to correct grammar on the interwebs without being condescending.
It should be mentioned that Linear Weights (which is just proper weighing of singles, doubles, triples and homers – the same stuff that’s used in wOBA) per LD + FB (including home runs) increases as pitcher ground ball rate increases. Using the same data source as Matthew, Gameday 2007-2009, and all pitchers with at least 500 balls in play, I get the following:
In case the picture doesn’t post, here is a link: http://www.hardballtimes.com/images/uploads/fline_thumb.jpg. The equation of the line is:
lw(ld+fb) = gb%*.27 + .06.
R-Squared = .236, P-Value < 2.2e -16
So that means for every 10% rise in ground ball rate, you'd expect the pitcher to give up .027 more runs per LD + FB. If the average pitcher allows a 45% ground ball rate and 300 FB + LD in a season, you'd expect him to give up roughly 54 runs per 300 balls in the air. If you give that same pitcher a 55% ground ball rate, you'd expect him to give up roughly 63 runs per 300 balls in the air. So that's a difference of 9 runs, which is basically 1 win. However, that 55% ground ball pitcher would give up fewer balls in the air (obviously) so the absolute difference in runs is going to be very close to zero, however, the extra runs per FB + LD is pretty large.
well if your looking at non home run balls then perhaps that is your answer. If both pitcher types give up 10 hard hit balls out of 50, but one type has 5 go for homers and you decide to exclude those, then those pitchers are going to have 5 hard hit balls out of 45 in your analysis. Whereas if the other pitcher only gives up 2 for homers and you exclude those, the it will appear that 8 out of 48 are hit hard. The second type appears to be hit harder in your analysis eventhough it is not true.
But in one of Matthew’s previous post in this series, he found that the rate of home runs per fly ball weren’t significantly different for GB pitchers or FB pitchers. What you’re suggesting shouldn’t apply here.
This could also be a function of the type of batter. Perhaps only the strongest of batters can muscle up a fly ball on the typical sinkerballer. Or because a pitcher is a sinkerballer most batters are trying to hit the ball for a single except only the biggest of powerhitters. Only those extreme guys maintain their powerswing against groundball pitchers, and therefore when they hit a flyball it travels more than average because it always travels more than average for those guys.
How are you handling infield flies?
Slugging percentage could go up even if the flyballs aren’t being hit harder. This is just conjecture, but could groundball pitchers perhaps give up more of those short flies that fall in front of outfielders? Maybe those aren’t even classified as flies, they could be more like looping line drives, which would nullify my point. Yet if they do give up more of these, it could lead to just more hits in general, which would increase slugging percentage. Would it be easy to look at if ISO increases in a similar fashion?
Scoring bias.
Scorers, after seeing ground ball after ground ball, are more likely to score high line drives as flyballs. This increases the slugging percentage on what are scored as fly balls because they are line drives
also what Dr Brian said
As everyone in the comments wants to make wild conjectures about why this is, I’ll join in, too.
Say that batted balls are scored correctly and consistently. There’s still a range of fly-ball trajectories and a range of line-drive trajectories. If you plot frequency of batted ball trajectory at a greater granularity than fly-line-ground you’ll probably see something like a normal distribution for most pitchers. No successful pitcher’s peak will be in the line-drive range. Successful fly-ball pitchers will have a peak high in the fly-ball range and ground-ball pitchers low in the ground-ball range.
If this is true that means the frequency of each trajectory for a ground-baller is declining through the LD and FB ranges, and increasing for a fly-baller. The ground-ball pitchers get more low flies, which are more like line drives (fewer outs and fewer homers) and more low liners (the same pattern must apply here if homers go down but slugging stays the same). Fly-ball pitchers, OTOH, trend higher in each bucket, yielding a lower BABIP and more HR.
The main reason the trajectory distribution *shouldn’t* be normal is that pitchers don’t throw the same pitch every time. So it should be a sum of normal distributions. A GB pitcher will throw mostly pitches centered in the GB range (sinker, slider); a FB pitcher will throw mostly pitches centered in the FB range (4-seamer); the more extreme a GB/FB guy he is, the more this is true.
Can you measure UZR by pitcher? Maybe there’s a correlation between OF UZR and the groundball rate of the pitcher.
The R-squared value on the results in all these posts are so low that it seems like there isn’t any relationship between ground ball rates and anything that has been looked at so far. On the four regressions shown, only one has a R-squared value over 0.023 (and that was 0.12). Basically it seems like the ground ball rates have no effect on what Matt has looked at so far and there is no relationship between ground ball rates and these other rates.