January 02, 2006
On the Ground and In the Air
Update 1/3/2006: Read through for a correction.
Cardinals Diaspora posts on the ground ball/fly ball ratios of the St. Louis starters and relievers. He finds the Cardinals like to hire pitchers who induce ground balls.
For a long time I've been bothered by ground ball/fly ball ratios, and since nothing is happening in major league baseball today, I thought I'd do a bit of research.
I don't believe there is a consistent view on what to measure when you're comparing ground balls and fly balls. STATS, Inc. would publish the ratio of all fly balls to all ground balls, leaving out line drives. Elias used to publish ground outs vs. fly outs, but I don't know if they included line outs in the fly outs. While the Ground/Fly ratio for the majors is around 1.2, the ratio of those on outs is close to 1. Here's a chart showing the percentage of each type of ball given up between 2002 and 2004 (bunts are not included, since a pitcher can't induce a bunt):
G = Ground Ball, F = Fly Ball, A = Air (Fly + Liners)
All |
Ground |
Fly |
Line |
G/F |
G/A |
2002 |
0.433036 |
0.354243 |
0.212721 |
1.222426 |
0.763781 |
2003 |
0.432865 |
0.342379 |
0.224756 |
1.264288 |
0.763249 |
2004 |
0.442254 |
0.370376 |
0.187371 |
1.194068 |
0.79293 |
|
|
|
|
|
|
Outs |
Ground |
Fly |
Line |
G/F |
G/A |
2002 |
0.460225 |
0.457944 |
0.081831 |
1.004982 |
0.852624 |
2003 |
0.466018 |
0.446196 |
0.087786 |
1.044424 |
0.872721 |
2004 |
0.463908 |
0.454623 |
0.08147 |
1.020424 |
0.865351 |
|
|
|
|
|
|
So it really matters if you include line drives. Pitchers in general get more balls in the air than on the ground. Notice two, that a fly ball is more likely to be turned into an out than a ground ball. About 35% of balls in play are fly balls, but about 45% of a pitcher's outs come on those same cans of corn.
So do ground ball/fly ball ratios really tell us anything? The data set is from 2002 to 2004, so I looked at all pitchers in that time frame with 400 innings pitched, and correlated their ERAs with the various ratios:
Pitchers with at least 400 IP, 2002-2004
Ratio | Correlation with ERA |
G/F | -0.137 |
GO/FO | -0.125 |
G/A | -0.155 |
G/AO | -0.114 |
There are three things you can glean from this chart:
- Ground/Fly ratios have very little to do with ERA.
- If anything, the correlation favors who give up a higher ratio of ground balls to fly balls, no matter how you measure them.
- The overall G/F ratio is a better measure than the number of outs on those balls.
I can certainly imagine that certain parks and certain defenses can change this calculation. But in general, we shouldn't worry about how balls are put into play against pitchers.
The data used in the correlation study is included in the extended entry.
Correction: I fixed point 2 above from lower to higher. Thanks to the people in the comments who pointed this out.
Pitcher | ERA | G per F | GO per FO | G per A | GO per AO |
Randy Johnson | 2.79 | 1.28 | 1.06 | 0.79 | 0.87 |
Pedro Martinez | 2.84 | 1.04 | 0.91 | 0.68 | 0.78 |
Johan Santana | 2.84 | 0.76 | 0.62 | 0.53 | 0.54 |
Jason Schmidt | 2.99 | 0.96 | 0.78 | 0.64 | 0.68 |
Tim Hudson | 3.04 | 2.41 | 2.03 | 1.35 | 1.58 |
Carlos Zambrano | 3.08 | 1.96 | 1.65 | 1.14 | 1.34 |
Mark Prior | 3.08 | 0.94 | 0.72 | 0.61 | 0.62 |
Curt Schilling | 3.17 | 1.10 | 0.90 | 0.70 | 0.74 |
Roy Oswalt | 3.19 | 1.38 | 1.17 | 0.82 | 0.97 |
Brandon Webb | 3.24 | 3.65 | 3.62 | 1.78 | 2.52 |
Kevin Brown | 3.32 | 2.10 | 1.84 | 1.18 | 1.46 |
Roy Halladay | 3.33 | 2.60 | 2.32 | 1.44 | 1.86 |
Barry Zito | 3.49 | 0.84 | 0.67 | 0.59 | 0.59 |
Kerry Wood | 3.50 | 1.03 | 0.86 | 0.67 | 0.75 |
Jacob Peavy | 3.53 | 1.10 | 0.88 | 0.71 | 0.74 |
Odalis Perez | 3.55 | 1.67 | 1.48 | 1.01 | 1.23 |
Greg Maddux | 3.56 | 1.98 | 1.88 | 1.11 | 1.46 |
Al Leiter | 3.56 | 1.07 | 0.83 | 0.70 | 0.70 |
Josh Beckett | 3.61 | 1.18 | 0.93 | 0.78 | 0.82 |
Tomokazu Ohka | 3.63 | 1.33 | 1.10 | 0.80 | 0.92 |
Tom Glavine | 3.64 | 1.37 | 1.13 | 0.85 | 0.91 |
Wade Miller | 3.66 | 1.20 | 1.03 | 0.76 | 0.89 |
Vicente Padilla | 3.68 | 1.52 | 1.37 | 0.94 | 1.10 |
Livan Hernandez | 3.71 | 1.40 | 1.17 | 0.84 | 0.96 |
Roger Clemens | 3.71 | 1.35 | 1.18 | 0.84 | 0.99 |
Woody Williams | 3.71 | 0.97 | 0.78 | 0.63 | 0.67 |
Mark Mulder | 3.72 | 1.95 | 1.70 | 1.16 | 1.39 |
Kip Wells | 3.72 | 1.73 | 1.54 | 1.03 | 1.30 |
Steve Trachsel | 3.73 | 1.12 | 0.94 | 0.72 | 0.80 |
Ben Sheets | 3.74 | 1.28 | 1.07 | 0.81 | 0.91 |
Andy Pettitte | 3.76 | 1.81 | 1.43 | 1.00 | 1.12 |
Matt Clement | 3.80 | 1.72 | 1.54 | 1.03 | 1.24 |
Joel Pineiro | 3.82 | 1.29 | 1.07 | 0.83 | 0.92 |
Doug Davis | 3.83 | 1.24 | 1.00 | 0.78 | 0.83 |
Randy Wolf | 3.85 | 0.96 | 0.80 | 0.63 | 0.67 |
Russ Ortiz | 3.85 | 1.18 | 1.01 | 0.74 | 0.85 |
Oliver Perez | 3.86 | 0.77 | 0.58 | 0.54 | 0.50 |
Mark Buehrle | 3.87 | 1.43 | 1.22 | 0.86 | 1.01 |
David Wells | 3.88 | 1.28 | 1.12 | 0.80 | 0.90 |
Jamie Moyer | 3.89 | 0.89 | 0.75 | 0.61 | 0.65 |
Bartolo Colon | 3.90 | 1.03 | 0.84 | 0.69 | 0.71 |
Kevin Millwood | 3.93 | 1.19 | 0.95 | 0.71 | 0.77 |
Mike Mussina | 3.96 | 1.13 | 0.97 | 0.72 | 0.82 |
Jake Westbrook | 3.97 | 2.90 | 2.72 | 1.50 | 2.00 |
Matt Morris | 3.97 | 1.56 | 1.41 | 0.99 | 1.20 |
Javier Vazquez | 3.97 | 0.91 | 0.74 | 0.61 | 0.64 |
Brad Penny | 3.98 | 1.22 | 1.04 | 0.79 | 0.87 |
Tim Wakefield | 3.98 | 1.05 | 0.92 | 0.73 | 0.82 |
Carl Pavano | 3.99 | 1.34 | 1.14 | 0.80 | 0.92 |
Brian Lawrence | 4.00 | 1.84 | 1.62 | 1.09 | 1.34 |
Jarrod Washburn | 4.01 | 0.74 | 0.57 | 0.51 | 0.50 |
Carsten Sabathia | 4.03 | 1.12 | 0.93 | 0.71 | 0.77 |
Carlos Silva | 4.04 | 1.79 | 1.54 | 1.08 | 1.30 |
Hideo Nomo | 4.05 | 0.89 | 0.77 | 0.57 | 0.64 |
Derek Lowe | 4.07 | 3.55 | 3.27 | 1.86 | 2.53 |
Ted Lilly | 4.09 | 0.78 | 0.67 | 0.52 | 0.58 |
Kirk Rueter | 4.11 | 1.50 | 1.29 | 0.89 | 1.02 |
Kelvim Escobar | 4.13 | 1.24 | 1.05 | 0.82 | 0.94 |
Brad Radke | 4.14 | 1.09 | 0.93 | 0.69 | 0.77 |
Elmer Dessens | 4.14 | 1.55 | 1.32 | 0.96 | 1.10 |
Ryan Franklin | 4.17 | 0.79 | 0.61 | 0.54 | 0.53 |
Mark Redman | 4.17 | 0.99 | 0.78 | 0.66 | 0.67 |
Miguel Batista | 4.21 | 1.82 | 1.52 | 1.08 | 1.30 |
Rodrigo Lopez | 4.22 | 1.17 | 0.98 | 0.75 | 0.83 |
Freddy Garcia | 4.24 | 1.10 | 0.94 | 0.72 | 0.82 |
Kazuhisa Ishii | 4.30 | 0.78 | 0.61 | 0.54 | 0.54 |
Jason Marquis | 4.34 | 1.76 | 1.55 | 1.03 | 1.25 |
Jeff Weaver | 4.38 | 1.14 | 0.94 | 0.70 | 0.81 |
Kenny Rogers | 4.39 | 1.45 | 1.30 | 0.82 | 1.01 |
Sidney Ponson | 4.40 | 1.63 | 1.45 | 1.01 | 1.22 |
Ramon Ortiz | 4.42 | 0.96 | 0.84 | 0.63 | 0.71 |
JohnT homson | 4.43 | 1.22 | 1.03 | 0.74 | 0.87 |
Adam Eaton | 4.44 | 1.01 | 0.89 | 0.67 | 0.77 |
John Lackey | 4.44 | 1.13 | 0.98 | 0.69 | 0.81 |
Esteban Loaiza | 4.57 | 1.24 | 1.05 | 0.77 | 0.88 |
Jeff Suppan | 4.58 | 1.39 | 1.21 | 0.84 | 1.00 |
Kris Benson | 4.59 | 1.03 | 0.88 | 0.64 | 0.75 |
Victor Zambrano | 4.60 | 1.27 | 1.05 | 0.81 | 0.89 |
Brett Tomko | 4.61 | 1.30 | 1.13 | 0.81 | 0.95 |
Paul Wilson | 4.62 | 1.18 | 1.00 | 0.73 | 0.83 |
Jason Johnson | 4.64 | 1.39 | 1.11 | 0.82 | 0.91 |
Jon Garland | 4.67 | 1.24 | 1.10 | 0.79 | 0.93 |
Brian Anderson | 4.68 | 1.01 | 0.89 | 0.62 | 0.75 |
Joe Kennedy | 4.68 | 1.10 | 0.90 | 0.72 | 0.79 |
Eric Milton | 4.70 | 0.61 | 0.49 | 0.44 | 0.44 |
Josh Fogg | 4.70 | 1.35 | 1.19 | 0.83 | 0.99 |
Kyle Lohse | 4.74 | 1.02 | 0.83 | 0.68 | 0.73 |
Mike Hampton | 4.74 | 1.85 | 1.47 | 1.02 | 1.16 |
Darrell May | 4.81 | 0.72 | 0.61 | 0.51 | 0.54 |
Glendon Rusch | 4.81 | 1.29 | 1.05 | 0.78 | 0.88 |
BrettMyers | 4.84 | 1.59 | 1.47 | 0.95 | 1.21 |
Cory Lidle | 4.85 | 1.66 | 1.46 | 0.95 | 1.15 |
Mike Maroth | 4.86 | 1.40 | 1.16 | 0.84 | 0.97 |
Mark Hendrickson | 4.88 | 1.30 | 1.11 | 0.79 | 0.90 |
Ismael Valdez | 5.00 | 0.98 | 0.85 | 0.63 | 0.74 |
Jason Jennings | 5.06 | 1.55 | 1.35 | 0.89 | 1.10 |
Aaron Sele | 5.20 | 1.08 | 0.87 | 0.68 | 0.74 |
Ryan Drese | 5.34 | 1.76 | 1.48 | 0.96 | 1.17 |
Tanyon Sturtze | 5.41 | 0.99 | 0.77 | 0.61 | 0.66 |
Steve Sparks | 5.51 | 1.25 | 1.09 | 0.80 | 0.95 |
Shawn Estes | 5.57 | 1.69 | 1.60 | 1.00 | 1.33 |
Posted by David Pinto at
04:35 PM
|
Pitchers
|
TrackBack (0)
Correct me if I'm wrong David, but doesn't a larger negative number for correlation mean that there's a smaller correlation? Or is it the absolute value of the number that matters?
But a decent pitcher can induce a fly or ground ball on a bunt. When a guy sets up to bunt (and most do so before the last second), a pitcher knows that a high pitch will likely induce a pop up whereas a sinking pitch will induce a ground ball. So unless I'm playing too many innings on my PS2, I'm thinking bunts should be included.
But I have to say I love the research you just did and compiled!
"If anything, the correlation favors who give up a lower ratio of ground balls to fly balls, no matter how you measure them."
I'm confused. Isn't negative correlation with ERA a good thing?
I agree with Adam and Rob, I would think a negative number would be a good thing in terms of ERA. Either way though, it is a small correlation as the numbers are close to zero.
What are we missing David?
I also agree with Devon, thanks for putting this together.
I checked MLB stats GO/AO for late inning relievers, & with a few exceptions, those who gave up more air outs
gave up a higher % of home runs. GO/AO may not matter as much for starters, but they seem to matter
for late inning relievers. A fly ball in a late inning, even
if caught, can ruin an otherwise nice day.
a negative correlation is DEFINATELY a good thing, and it makes sense: a pitcher who makes batters put the ball on the ground more often SHOULD (can't say anything's a given in baseball) be more successful than a pitcher who gives up flies. and even though ERA is reliant on team defense, and some luck, it is generally a good measure of how successful a pitcher is -- At least, you can't have a good ERA if you totally suck.
why would anyone prefer a pitcher who gave up mostly flies to one who gave up mostly grounders?
If a team had a really crap bunch of infielders, and a fine set of outfielders, they may prefer the FB type pitcher. But how common is that?
This is great analysis, David...which of course, begs more questions. Line drives are clearly the primary killer of pitchers, with the worst pct in play/pct out ratio by far. Yet, fly balls have the highest pct in play/pct out ratio. Is there any way to examine individual pitchers over time, and see how effectively they can control their FB/LB ratio? It seems that pitchers are often consistent in their GB/AB ratio from season to season, which appears to take the luck factor out of the equation. But if pitchers can consistently control their FB/LB ratio, that would actually seem to be the best predictor of ERA. Perhaps GFB/LB would be the ideal pitcher measure?