Where are David Price’s Grounders?
Last week R.J. looked at David Price’s improving K/BB ratio. Another important part of the Price story I wanted to look at is ground balls. Coming up through the minors one of the things that made Price so exciting was his combination of strikeouts and ground balls. Below are his GB/BIP numbers by year and level. The major league numbers are Baseball Info Solutions data from here at FanGraphs. The minor league batted ball data was taken from StatCorner, which gets it from Major League Baseball Advanced Media.
n is number of balls in play +------+-------+-----+--------+ | Year | Level | n | GB/BIP | +------+-------+-----+--------+ | 2008 | A+ | 90 | 0.50 | | 2008 | AA | 149 | 0.58 | | 2008 | AAA | 56 | 0.54 | | 2008 | MLB*| 40 | 0.50 | | 2009 | AAA | 93 | 0.41 | | 2009 | MLB | 219 | 0.36 | +------+-------+-----+--------+ * includes time as starter and reliever
At all levels in 2008 he had great ground ball numbers, even in his brief exposure to the Majors. But in 2009, in AAA and the majors, his ground ball rate has plummeted.
For the most part Price throws two pitches, a slider and a fastball (as a starter he also throws a changeup and very rarely a curveball, but over 90% of his pitches are fastballs or sliders). Here are the ground ball rates on those two pitches during his major league career. Pitch-by-pitch data is not available for the minor leagues.
+----------+------+--------+ | Pitch | Year | GB/BIP | +----------+------+--------+ | Slider | 2008 | 0.62 | | Fastball | 2008 | 0.49 | | Slider | 2009 | 0.50 | | Fastball | 2009 | 0.31 | +----------+------+--------+
I am not 100% sure why we see the big drop in ground balls from both of his pitches. He is locating them roughly in the same part of the zone this year as last. One possible reason is that his fastball has about an inch and a half more ‘rise’ this year compared to last year (9.6 in versus 8 in). This could result in fewer grounders. He throws both pitches slower since he is starting this year, which could have something to do with it.
Another interesting aspect is that he throws both a four- and two-seam fastball. I think almost all of his fastballs out of the pen last year were four-seamers, although I am not 100% sure. This year he started off throwing mostly four-seam fastballs, over 90% of his fastballs were four-seamers. But recently he has been throwing the two-seam fastball more often. In his last two starts, last night and August 5th, about 30% of his fastballs were two-seamers.
Maybe the Rays have noticed the lack of ground balls and are looking for him to throw more two-seamers, which generally are more of a ground ball pitch than four-seamers. I am not sure what his breakdown was in the minors, although at Vanderbilt he said 80% of his fastballs were four-seamers.
It is an interesting trend to keep an eye on.
Dave Allen's other baseball work can be found at Baseball Analysts.
I thought I was the only one to notice! A bit of a worry going forward. I am hoping he can turn into an Erik Bedard (but not a total wimp) type. Good GB & K rate. He needs to keep developing his change, it will likely never be above average, but he needs to change in speed… When I watched him pitch against the Jays (a start he had absolutely no stuff) his fastball sat around 89-91, wonder if this was the start of throwing the 2-seamer? I thought he was labouring, didn’t really notice the grip or much down and away movement from righties. He normally has a bit of arm-side run to his fastball as is though.
He just needs to throw his slider and other offspeed pitches more. A two seamer helps but if almost 90% of his pitches were fastballs like he did against the Angels, there is no reason to expect them not to sit on it and kill it for base hits.
The sample size of 14.0 IP in the MLB in 2008 makes any analysis on GB% by pitch type unreliable.
This is an interesting question, and Mr. Cameron does not assert that the theory his 1.6 increased “rise” on the fastball is necessarily correct, but analysis on pitch type (which divides the small sample further by categorizing int by pitches) from a sample of 14.0 IP is meaningless.
I wrote the article, not Dave Cameron. It is confusing with all the Daves around.
Good point about the small sample size. You are right there is very little we can infer from those 14 innings. But generally the expected groundball rate off a fastball varies inversely with the ‘rise’ of the fastball. I will dig up the reference for that, and if I cannot find one I have done the work and will post it here or at Baseball Analysts.
My bad, it was you Senior Allen.
Agreed on the ‘rise’ of the fastball and it’s correlation to GB%. Your article on Affeldt displayed that beautifully if I recall correctly.
One question, ‘rise’ (or really just vertical movement generally) in PitchFX is considered to be the vertical distance (in inches) that a pitch deviates from the expected trajectory at a given velocity with gravity included, correct.
So a FB with positive vertical movement, i.e. ‘rise’ (which I believe you refer to by placing it in quotes) before it crosses a certain threshold actually means the ball goes more “straight”, right?
To clarify, by correlation in my previous post I meant negative correlation.
One question, ‘rise’ (or really just vertical movement generally) in PitchFX is considered to be the vertical distance (in inches) that a pitch deviates from the expected trajectory at a given velocity with gravity included, correct.
Exactly. So I say ‘rise’ for pitches with positive vertical movement, because they do not actually rise as the move to the plate. But they drop less than you would expect due to gravity.
It seems to go down as he rises through the ranks.
I think the answer is pretty simple: Sample size, with a sprinkle of MLE thrown in. Ground balls are tougher to come by in MLB which accounts for the drop from his MiLB numbers. And then trying to compare his raw small sample size 2008 and 2009 GB rates without some regression is bound to cause a bit of head scratching. In a nutshell, I don’t think we ever really knew with certainty what his MLB GB% was going to be, and shouldn’t really be surprised by his numbers this season, especially if a little variance is added to the mix.
Your theory doesn’t withstand scrutiny. David Price is throwing less GBs in 2009, and MLE doesn’t explain it. The sample sizes are pretty small, but these relationships are almost statistically significant.
[in R]
> MLB_08 MLB_09 MLB MLB
[,1] [,2]
MLB_08 20 20
MLB_09 79 140
> fisher.test(MLB)
Fisher’s Exact Test for Count Data
data: MLB
p-value = 0.1120
alternative hypothesis: true odds ratio is not equal to 1
95 percent confidence interval:
0.8464159 3.6972463
sample estimates:
odds ratio
1.768135
Our p-value from this Fisher’s exact test comparing GBs to non-GBs is almost less than .1 (which isn’t the generally used scientific p-value, that’s usually .05, but many companies/government institutions use .1 including the US Census Bureau). We don’t quite have a statistically significant difference, but it’s so close we have to consider this a meaningful relationship in this real world of incomplete information we live in.
Similarly, the AAA relationship is almost statistically significant.
> AAA
[,1] [,2]
AAA_08 30 26
AAA_09 38 55
> fisher.test(AAA)
Fisher’s Exact Test for Count Data
data: AAA
p-value = 0.1742
alternative hypothesis: true odds ratio is not equal to 1
95 percent confidence interval:
0.8112947 3.4407261
sample estimates:
odds ratio
1.664258
Here it is .1742, which is also not quite significant but still suggests that there is something there.
So to reiterate, no there isn’t a simple explanation, and if there was it certainly is not MLE.
that looked prettier when I pasted it, R does not paste well into a text box, the whitespace gets fucked up. Oh well, people should still be able to follow it.