Baseball Musings
Baseball Musings
January 16, 2006
Ground and Air Averages

A little over a week ago this site looked at who was responsible for getting ground balls. A reader wondered if we could plug batting stats into this method, so here it is. Let me remind you of the methodology:

To study this, I selected a group of pitchers that gathered 400 IP from 2002-2004. There were 101 pitchers in the group, and I divided them into quartiles on the probability of a ground ball. Quartile 1 is the group with the lowest probability of a ground ball, quartile 4 the highest. I also selected batters with 1000 plate appearances in that time frame. There were 243 batters in the study, also divided into quartiles on the same statistic.

Let's start with batting average. First here are the overall averages for the quartiles:

Batting Average by Quartile (1=least grounders, 4=most grounders)
QuartileVs. PitchersBatters
1.259.274
2.261.279
3.269.277
4.265.278

Remember, this study is looking at regulars on both sides of the ball, therefore we expect these groups to be better than average, hence the differences in batting average between batters and pitchers. Next, I pitted each pitcher quartile vs. each batter quartile. The 1-1 cell are pits pitchers and batters who tend to get lots of balls in the air vs. each other, while the 4-4 cell are pitchers and batters who keep the ball on the ground:

Pitcher Quartiles vs. Batter Quartiles, Batting Average
PitchersBatters
Quartiles 1234
1.258.259.269.272
2.266.275.271.270
3.277.294.284.281
4.274.283.275.268

It looks like quartile two hitters (slightly fewer ground balls), are the best for generating hits, while the quartile 1 pitchers (fewer ground balls) are best for preventing hits. Maybe these type 2 hitters represent a more versatile group, one that can adjust to the type of pitcher on the mound.

Now let's take a look at slugging percentage. First the average for the quartiles:

Slugging Percentage by Quartile (1=least grounders, 4=most grounders)
QuartileVs. PitchersBatters
1.430.482
2.417.469
3.423.443
4.405.409

Here, slugging percentage appears to have a lot more to do with the tendancies of the batter than the pitcher. Now let's look at quartile vs. quartile:

Pitcher Quartiles vs. Batter Quartiles, Slugging Percentage
PitchersBatters
Quartiles 1234
1.463.443.441.410
2.464.462.432.389
3.478.492.455.407
4.458.456.424.370

Again, slugging averages go down as you go from low grounder to high grounder batters. What's really fascinating is pitching quartile 3, which has the highest slugging percentages of all the pitchers in three of the four batting groups. Why is this?

It might be that this group are the ground ball pitchers who make mistakes. Bruce Hurst comes to mind. If he hung a pitch, it just got hammered. Pitching quartiles 1 & 2 are successful with balls in the air. Pitchers in quartile 4 don't hang pitches. But quartile three is where you find pitchers who make mistakes. I'll have to figure out what pitchers are in each quartile and see if that idea holds up.

Correction: Fixed the caption on the last table.


Posted by David Pinto at 05:38 PM | Statistics | TrackBack (0)
Comments

Nice follow-up study! Interesting to see the difference between slugging and batting averages. Also interesting that there is little reason for a hitter to try to hit grounders---the slight boost to BA doesn't compensate for the huge loss in SLG. Your hypothesis about q3 pitchers seems very plausible. Is there a study that could be done to investigate? Seems like it'd be very hard to evaluate... Another thing I wonder: what kind of BB rates do we see for the various pitcher quartiles? Do 1 & 4 pitchers tend to give up more walks?

Posted by: Jason at January 17, 2006 08:44 AM

David,
as context to your charts, the ML average from 2002-2004 was .264 BA and .422 slg. [You didn't say that you eliminated pitcher vs pitcher ABs; assuming you didn't, that's also part of the reason the pitcher quartiles show lower BA and slg pct than the hitters]. Do you think it makes sense to remove pitcher vs pitcher results? My guess is that as a group, pitchers as batters are extreme groundball hitters, as well as having low batting averages and slugging percentages. You might see some national league and american league pitchers change quartiles if you adjusted for this, since (if I'm right) national league pitchers, by facing pitchers as batters more often, will have their "true" groundball tendencies exaggerated in comparison to american league pitchers. This could change the results when you match the quartiles.

Finally, since I think your goal is to isolate the groundball/flyball tendency alone, do you see any bias in the distribution of lefties/righies in the quartiles? traditional platoon effects might not fully cancel out across these quartiles.

Posted by: joe arthur at January 17, 2006 10:21 AM
Post a comment









Remember personal info?