THE BOOK--Playing The Percentages In Baseball

"You pull a ball on the ground, you make an out.  You go the other way, you get a hit. You pull a ball in the air, you get a hit.  You go the other way, you make an out.”

Is that right?  I thought pulled balls are more likely to be a hit either way.  What I think is true is that pulled balls tend to GBs, while balls hit the opposite way tend to be air balls.

What I think is true is that pulled balls tend to GBs, while balls hit the opposite way tend to be air balls.

This is correct.  Pulled balls are hit harder because the bat has more time to accelerate.  However, because of the downward angle of the bat, a batter who hits the top of the ball will tend to pull it, and a batter who hits the bottom of the ball will tend to hit it to the opposite field.

This was best explained by Matt Lentzner in one of my favorite articles ever, here:
http://www.hardballtimes.com/main/article/why-flies-go-one-way-and-grounders-go-the-other/

Btw, the HITf/x data from April 2009 confirms these theories.

Matt Lentzner wrote an article for THT back in (I think) 2007 in which he addressed the issue of mis-hit balls.  Because the bat is usually titled with the barrel below the knob during contact, balls that are severely undercut by the bat (i.e., bat hits well below the center of the ball) tend to go the opposite way and are pop-ups while those that are severely overcut tend to be pulled and are grounders.  Matt explained it in the article quite elegantly and I won’t try to repeat the argument here.  But his result is not inconsistent with

“You pull a ball on the ground, you make an out.  You go the other way, you get a hit. You pull a ball in the air, you get a hit.  You go the other way, you make an out.”

Interesting stuff. When I saw Pineda and Strasburg up there my first thought was that logically, the Pull% should correlate with Fastball velocity. I cross-referenced Jeff’s list with Fangraphs’ average FBv leaders and this is what I found: (Rank in parentheses of FBv in 2011 among 117 qualified starters)

Pitcher Pull % (FBv Rank)

Pineda 21% (1st)
Lincecum 24% (16th)
Weaver 24% (72nd)
Johnson 24% (6th)
Verlander 26% (2nd)
2010 Strasburg 27% (1st in 2010)
Halladay 29% (48th)
Felix 30% (11th)
Haren 33% (82nd)
Shields 35% (52nd)
Beckett 35% (20th)
Cahill 35% (92nd)

There are some of the same names on the top of both lists, but not as strong of a connection as I thought. Anyone have any ideas how pitchers like Cahill and Weaver are getting similar results to pitchers with FBv 5 mph faster?

Because I am too slow editing my own comments, Mike beat me to it!

Guy/Mike: I don’t think either of you are disputing me.

On GB by LHH, the out rate is higher on balls hit to 2B than hit to SS.  And the out rate is higher on balls hit to 1B than hit to 3B.

On GB by RHH, the out rate is about the same if hit to 1B or 3B.  And while the out rate is higher still with 2B than SS, the gap isn’t as large.

I have this data lying around, so I’m not sure exactly from where or when it is, but:

LHH pull: 75% out
LHH opposite: 63% out

RHH pull: 70% out
RHH opposite: 70% out

So, pull the ball, and you make an out 72% of the GB, and go opposite and you make an out 67% of the GB.

Tango/6, I wasn’t necessarily disputing you.  I was providing data to back up the specific claim that Guy made.

When I look at the HITf/x data from April 2009 for ground balls (vertical launch angle < 7 degrees), I get the following results.

LHH pull: 73% out (2133 balls)
LHH opposite: 67% out (830 balls)

RHH pull: 72% out (948 balls)
RHH opposite: 75% out (2842 balls)

A much bigger sample size might trump the accuracy of the HITf/x trajectory data.

Mike#7:  RHH pulled the ball only about one quarter of the time?

Guy, swap the two.

Swap the 2 including the out percentages? Is it RHH Pull: 75% out or 72% out?

Excellent point!  I’m hoping it’s swapping both, as that would be more consistent with what I have.

should have said

RHH opposite: 72% out (948 balls)
RHH pull: 75% out (2842 balls)

Mike: excellent.  So, that is fairly consistent with what I have, but just less extreme at the handedness level.

Mike is saying overall: 74% out rate on pulls, and 70% going the other way.

I’m saying 72% on pulls, and 67% going the other way.

Same thing.  My source by the way was Gameday, though I don’t remember the year.

Interesting data.  I can see why pulled balls become outs more often—the defense is aligned for a pulled ball, since that is the most common outcome.  But I’m surpised LHH don’t do a little better when pulling, since the 1B often has to hold the runner.  Perhaps the shorter throw to first cancels that out. 

I also wonder if a few extreme players perhaps drive the LHH results.  I’d guess Ichiro accounts for a non-trivial share of opposite GBs by LHH.  It seems like there are more LHH than RHH who specialize in slapping the ball the other way (Boggs, Gwynn, Ichiro), although that may just be a perception.

Mike and Tango - How are you both defining pulled ground balls?  Mike, did you use the horizontal angle from Hit Fx?  Did you consider every ball to the third base side of dead center a pulled ball by a right hander and vice versa, or did you use some other criterium?  Tom, did you decide by who ultimately fielded the ball or did you use the Gameday hit location data?  If the former how did you decide on hits fielded by the CF or did you leave them out?  I am asking because I am getting a much narrower difference of 75.1% outs on pulled ground balls and 73.6% outs on GBs to the opposite field using Gameday hit locations and dividing the field in half.

Peter: I used whatever classifications are in gameday.  I removed bunts.  I included MLB and Minors.  I used the hit location data (but adjusted, since not every park has home plate in the same spot in Gameday).  I divided at 2B bag.

***

I suppose we can technically argue that the dividing line should be somewhere at 10(*) degrees offset from 2B, since the bat, at the point of contact, on average is not perpendicular to the line from 2B to home plate, but rather some point between shortstop and the bag.

(*) Number for illustration purposes only.

Mike, did you use the horizontal angle from Hit Fx?  Did you consider every ball to the third base side of dead center a pulled ball by a right hander and vice versa

yes

The article originally cited in this thread I believe used a Pulled, Up the middle, Opposite division which was more what I was thinking of when I asked the question about where the dividing line was.

Going back to the original article, it does seem that looking at percentage of balls pulled could be valuable, and perhaps more stable than HR/PA or HR/FB.  You could look at the percentage of a pitcher’s airballs (FB or LD) that are fielded by the OF on the pull side (whether hit or out).  That could prove to be a pretty good indicator of HR prevention skill.  And it involves a minimal amount of subjective judgment by the stringers (just a few marginal LD vs. hard-hit GB calls).

I’ve asked this question before in other threads, but it certainly relates to the Lentzner article and this topic: 

Could I use this info to help justify a five-infielder defense at the high school level? 

Using this data, how would I position the outfielders, if I were to employ one? (against both LH hitters and RH, and depending upon whether I am throwing a lefty or righty). 

(My theory, which I hope to test, is that the new aluminum bat regulations have made the five-infielder defense the optimal defensive alignment in low-level varsity baseball.  It’s really hard for most kids to hit the ball to the opposite field with authority).