More fun with PITCHf/x

So I came up with an algorithm to try to identify what pitches pitchers are throwing.  You can read more about it at my blog if you are interested.  Anyway, I thought I would highlight a few of the Brewer pitchers here.  It is a bit of a work in progress but it is getting there.  All plots show the movement of the pitches not the location.


So the algorithm isn't perfect and you can see it with Gallardo.  The pitches listed as a splitter is a curve but the relatively large horizontal break compared to the vertical break the algorithm is messing up and calling it a splitter.  The real question is where is the change up in all of that?  It appears that a few of those could sliders could be change ups but with such a low percentage it is getting grouped in with the sliders.  I wonder if the relatively small horizontal break of his fastball is why he has such good command?


Nothing too surprising here.  Almost all fastballs and sliders.  The algorithm thinks he has thrown a couple of changes in there.  Not sure if that is the case or not.


Again, the clustering algorithm is grouping Cappy's slider and change together.  The real change ups are just below the fastballs.  You can see why that pitch should be a good pitch for him as it resembles the fastballs except for the speed.


Shouse's fastball is definitely a two seem variety and has some huge horizontal break.  You can see how tough he is on lefties as the ball breaks from their thighs to just over the plate.  His offspeed pitch is listed as a splitter (can anyone confirm that?) that rises slightly and moves the other direction than the fastball.

Anyway, it is time for bed here.  Maybe I will post some more tomorrow.