Over at the main page, bgh recently asked How good is Joe Kelly's Sinker? Comparing vertical and horizontal movement, spin, rotation and velocity to those of proven groundball-specialists, he makes the case that Kelly's sinker is in good company and should go on to generate plenty of groundballs in the future. While I completely agree with his assessment, I kept wondering what actually makes a sinker produce groundballs aside from, well, sink. I looked into the matter and would like to share some of the results here.
Preliminary note: This analysis is based on 75 righthanded pitchers who are throwing sinkers at least 20% of the time. This has some implications when interpreting this data. As brackenthebox points out in the comment section of above linked article, pitchers who rely on the sinker, or at least use it frequently, are also more likely to adopt an approach that is aimed at generating groundballs in general. Therefore, the findings presented here should not be compared to non-sinkerballers or other pitches. However, they can shed some light on the question why some sinkers induce significantly higher rates of groundballs than others.
Expecting that sink is indeed the driving factor behind gb%, I looked at vertical movement first and found a distribution that ranges from -12.47 to -29.13 within my sample . The explanation for differences in vertical movement is rather simple:
The coefficient of determination for spin axis and vMov is 0.801. The more the spin axis gets away from a straight backspin (180°), the more vertical movement. Pretty much what you'd expect. The dot in the lower right corner is Justin Masterson. Besides spin axis, the other noteworthy factors in regards to vertical movement are velocity (0.203) and rotation (0.428):
Less rotation equals more downward movement, while increased pitch speed has a tendency of reducing it, although much less than one might expect. And in case you're wondering, the guy in danger of falling off the wrong side of that velocity chart is Livan Hernandez. The two outliers in the rpm department are Aaron Cook and Jhoulys Chacin, whose sinkers show characteristics much closer to that of a splitter. For some provisional results: The degree of the spin axis seems to be the main factor in generating downward movement, with some assistance from (less) rotation and (less) velocity, in that order.
Now to the actual question: What determines the groundball rate of the common sinker? Makes sense to ckeck the correlation between gb% and vertical movement first:
Still no surprises. Vertical movement correlates well with gb% (0.475), indicating that it is indeed the main cause for higher gb%. Moving on.
While velocity does tend to reduce the vertical movement of sinkers, there's almost no connection between velocity and gb% (0.005). A simple explanation for this could be that while the reduced sink induces less groundballs, the higher velocity in itself generates weaker contact and, in turn, more groundballs. Taking into account that velocity does correlate with whiff% (0.202), the obvious lesson from this would be that a higher velocity sinker is preferable, since it doesn't reduce gb% while at the same time generating more swings and misses.
Interestingly, horizontal movement does not seem to factor into the production of groundballs at all (0.074). In one way, this makes sense, since a groundball is the result of a batter hitting the ball with the lower half of his bat and horizontal movement does nothing to put the ball there. On the other hand, I would have expected any kind of movement to generate weaker contact, but this seemingly does not manifest itself in the gb% of sinkers.
So, spin axis and rpm are mostly responsible for vertical movement, which is the key factor in generating groundballs and Boardwalk Empire has taught me to always cut out the middle man:
RPM and gb% correlate well (0.301), spin axis and gb% even better than vertical movement and gb%, albeit only slightly (0.496).
To quickly summarize: Most of this comes as expected. Vertical movement seems to be the main reason for higher gb%, with spin axis and rpm being mostly responsible for that downward movement. Somewhat surprising, to me, is the lack of effect that velocity and horizontal movement seem to have on gb%. As a general rule, vertical movement and/or a combination of its underlying factors - spin axis and rpm - seem to be a reliable indicator of expected gb%. Velocity should not be dismissed entirely, since its negative effect on vertical movement does not seem to influence gb% and is offset by a higher whiff%. All that being said, there are more factors which were not considered here. Location, for one thing, obviously plays a big role in this, as well as usage of the pitch (frequency and situation).
For the case of Joe Kelly's sinker, his gb% stands at a paltry 32.14 right now. Given the spin axis of 252, an rpm of 1995 and vertical movement of -21.08, I think one could reasonably expect the gb% of his sinker to be much closer to the 55%-60% range going forward.
If you're interested in all the correlations, here's a chart.