When I posted about the Mariners' 5th starter candidates, a discussion about ground balls broke out and I added a follow-up post based on a quick comparison look between ground balls and fly balls. That post busted a levee of ideas and I ended up delving into tons of data from 2007-9 to look at how different variables interacted with a pitcher's ground ball rate.
I ended up posting a series of articles on the information gleaned on FanGraphs and the thought occurred to me that some of you may not have seen it. I don't like to cross-post material, but since it's been a week and I started it here originally I figured it was worth opening up for discussion.
Ground balls vs. fly balls
Ground balls and home runs
Ground balls and slugging rates
Ground balls and walks and strikeouts
Ground balls and runs scored
I encourage you to go read through them for the graphs at least, but here are some key excerpts:
The results were that the average ground ball generated 0.04 runs and caused 0.80 outs while the average ball in air generated 0.23 runs and caused just 0.62 outs. On a runs-per-out basis, balls hit into the air created almost 7.5 times as much offense as balls kept on the ground did.
What constitutes a line drive is somewhat fuzzy and open to subjective bias. However, even with ignoring line drives, fly balls and pop outs by themselves generate an average of about 0.1 runs and cause 0.79 outs. That rate is still about three times more offensive than the average ball hit on the ground.
The best rule of thumb I can state from this look is that a pitcher’s ground ball rate has no impact on his various rates of yielding home runs and what impact there is might actually be negative.
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.
The expected difference between the two biggest reasonable extremes in ground ball rate (30% to 60%) amounts to fewer than five strikeouts or walks over 200 innings pitched.
To re-state, all data came from MLB 2007-9 seasons and pitchers were restricted to those with at least 250 total balls in play over that entire span.
0 recs | 33 comments
"busted a levee of ideas"
Which resulted in catastrophic floods in the Lower Ninth Ward of Ignorance.
marc w - February 24, 2010
Note that, in the GB vs. RA/27 graph, there's a perception difference between R^2 = 0.0218 and R = 0.1476
Given that groundball pitchers allow more unearned runs due to the frequency of errors on grounders, do you think it’s at all worth considering that a GB pitcher’s FIP and a FB pitcher’s FIP can be a little misleading?
Jeff Sullivan - February 24, 2010
A little, yes.
Matthew - February 24, 2010
Are errors a good way to estimate deviation of R/9 from FIP?
At first the increase in errors could cause you to think that GB pitchers have a larger deviation since they appear to put more pressure on the defense to make plays. When I think about this more though, if an IF misses a ball its probably called an error. When an OF can’t get to a ball it probably is not an error and therefore FB pitchers might not see a significantly greater R/9-FIP split than GB pitchers. Is this true?
Edgar for Pres - February 24, 2010
If you're saying what I think you're saying, then yes
Groundballs often lead to errors. Fly balls seldom lead to errors. Groundball pitchers will have more errors behind them, so groundball pitchers will allow more unearned runs, so groundball pitchers will show a greater difference between RA and FIP.
Jeff Sullivan - February 24, 2010
Let me try to rephrase
I agree the GB pitcher will have more errors and therefore his RA-ERA will be greater than the FB pitcher. Are we sure that his RA-FIP will be higher though? Errors don’t measure all defensive mistakes. Players who make mistakes on flyballs are probably less likely to have errors. I don’t think its clear to me that RA – FIP will be higher for a groundball pitcher because you are saying that outfielders don’t make more mistakes as the number of flyballs increases. I’m guessing RA – FIP will be probably roughly the same for either GB or FB pitchers because the defense makes mistakes on both FB and GB so the effect of an increased number of either will probably be small.
Edgar for Pres - February 24, 2010
I think what I'm saying is that groundball pitchers will show a greater difference between RA and ERA.
Not RA – FIP. Errors aren’t related to RA – FIP very strongly although they are related to RA – ERA.
Edgar for Pres - February 24, 2010
Based on pitchers with 200+ IP, 2007-2009
Correlation® between GB% and RA-FIP: 0.18 (slope of 1.45)
Correlation® between GB% and ERA-FIP: 0.12 (slope of 0.86)
Correlation® between GB% and RA-ERA: 0.22 (slope of 0.59)
Jeff Sullivan - February 24, 2010
After playing around a little more, I am not really finding an effect
Whatever. If there’s anything here, it’s small.
Jeff Sullivan - February 24, 2010
Yeah I just tried to play around with it too with GB/FB ratio to try to get around LD% fluctuations
And couldn’t find anything close to the stuff you were showing so I figured I might have done something wrong.
Edgar for Pres - February 24, 2010
Cool. I'll need to rethink this a little
Just one point, I think to compare those slopes all those should be rescaled to be on the R/9 basis. (FIP*1.09 right?) Maybe that’s not a useful thing to do…More thinking required.
Edgar for Pres - February 24, 2010
Ehh, I gave up thinking about it
Not sufficiently interesting or meaningful to investigate further.
Jeff Sullivan - February 24, 2010
Yeah I kind of agree
Nothing was coming quickly to me and I agree there isn’t much point in looking into it more. I just wanted to do a reality check because something seemed like it might have been off. I’m not sure if you were originally right but I’m fine with both of us walking away and doing something more important.
Edgar for Pres - February 24, 2010
I stand by those correlations
But those are r’s, not R^2’s, and the effect is so weak that there’s no reason to further complicate what we already have.
Jeff Sullivan - February 24, 2010
Yeah I believe your numbers. I was using a crappier data set
Edgar for Pres - February 24, 2010
Actually I got very similar r values using GB/FB
I was using R^2. Good call.
Edgar for Pres - February 24, 2010
ground ball fly ball
How do sacrifice flies count? This obviously would affect runs scored.
undeterredphil - February 24, 2010
A sacrifice fly?
abender20 - February 24, 2010
Count as a ground ball at home, fly ball on the road
seattlebruin - February 24, 2010
It's the second time I read this
and I laughed both times.
morrow - February 24, 2010
I never comment on Fangraphs, but this post affords me the opportunity to compliment your work.
That was a fantastic series of articles on an extremely interesting subject. Very well written as well, thank you.
Kermit. - February 24, 2010
That last block quote...
When you said that, is that the difference in run expectancy?
To me it sounds like the difference between an extreme FB pitcher and extreme GB pitcher is only 5 strikeouts over 200 innings. Which doesn’t make sense to me if a fly ball creates 7.5 times as much offense as a ground ball.
Feel like I am missing something here…
bhsmarine - February 24, 2010
A fly ball and a strikeout are independent of eachother
vivaelpujols - February 24, 2010
People had speculated that groundball pitchers might allow more walks or strike out fewer batters than fly ball pitchers
That bullet point addresses this.
Jeff Sullivan - February 24, 2010
Understand that
I am just confused how a possible 30% difference in ground ball rate amounts to fewer then five strikeouts. If they are independent then how do they amount to each other?
bhsmarine - February 24, 2010
Ooops
Missed the reply button.
Jeff- Gotcha, that was the link I was missing between the two. Thanks.
bhsmarine - February 24, 2010
Hey Matthew
Do you think you can run some regressions on the following things?
-Strikeout% on Runs/27
-Popup% on Runs/27
-Walk% on Runs/27
-LD% on Runs/27
Groundball% appears to account for about .5 runs per 27 if you look at the most extreme values. I’m just curious as to how that stacks up with some of the other skills. Thanks a lot.
vivaelpujols - February 24, 2010
I'm not Matthew but watch me try
K% on Runs/27: Y = -12.93x +6.89, R = -0.65
BB% on Runs/27: Y = 8.20x + 3.95, R = 0.19
LD% on Runs/27: Y = 23.93x + 1.24, R = 0.43
IFFB% on Runs/27: Y = -3.81x + 4.88, R = -0.07
Jeff Sullivan - February 24, 2010
Thanks Jeff
So Groundball percentage has a slope of -2.3 and an R of .14. So IFFB% has a slightly greater slope, but basically no correlation. The rest of the stats have a much bigger impact (especially LD%, which is like huge. Every 5% increase in LD rate is an extra 1.2 runs per 9, if I’m understanding that right – wow).
vivaelpujols - February 24, 2010
That's what happens when you allow a lot of a batted ball type that drops in 70% of the time!
Jeff Sullivan - February 24, 2010
By the way, that's LD as a function of PA, not BIP
Jeff Sullivan - February 24, 2010
Oh, that's why then
vivaelpujols - February 24, 2010
So basically strikeouts are devastating
I probably missed this somewhere, but what’s the FB% on RA/27?
OlSalty - February 24, 2010
You must Login with your SB Nation account and be a member of Lookout Landing to post a comment.