Skip to content

Pitchers and Ground Ball Quality

September 24, 2010

4 posts in 2 days. Look at PAH9 go. The following post is indirectly related to my Brendan Ryan post. You also should check out Andy’s pet peeve and my top 7 Cardinal prospects.

One of the common criticisms (especially among Cardinal fans) of DIPS pitching stats is that all batted balls are not created equal. Specifically, all GBs are not equal, all FBs are not equal etc. With that thought in mind I wanted to compare a couple of Cardinal pitchers to see if there was a discernable difference in their GBs. To level the playing field I only looked at RHB when Brendan Ryan was playing. I looked at Out+Error rate, making the assumption that the pitcher had no control over the error part. The following table summarizes the results across all hit angles (GBs only)

Out+Error Rate BIP
Carpenter 0.820 311
Wainwright 0.799 328
Lohse 0.731 186

And then across the SS area of responsibility (since that was who we held constant)

Out+Error Rate BIP
Carpenter 0.870 154
Wainwright 0.887 151
Lohse 0.818 88

And in chart form

So what can we attribute the differences to?

Ground ball quality – I would guess that harder hit ground balls would be more likely to make it up the middle (-7.5 in the chart) and through the hole (-27.5). The anecdotal evidence in the data above seems to agree. I’d guess that Lohse gives up the hardest hit balls of the 3.

Defense – Yes Brendan Ryan was in the field, but that isn’t to say that he played identically (both in reaction time and positioning) behind all 3 of these guys. As samples increase this effect would likely decrease.

Park effects – Infield speed isn’t constant across all parks (can adjust for these, but haven’t)

Stringer/Scorer Bias – Are these all groundballs? What is the difference between GBs and LDs? Is the hit location recorded accurately? Is there Hit/Error bias?

Luck – Bad hops, deflections etc.

The real question is what weight you put on each of those factors. I’m not sure we’ll get at the answer to that until we get Field F/X data (if we get field f/x data). For now I’d hesitate to weight the first one (which is what would be ideal to measure) as any more than 50% of the difference. There’s just too much other stuff that could be at play.

Comments are closed.

%d bloggers like this: