New Arms of the Week - August 10-16

by Harry Pavlidis on Aug 17, 2009 7:23 PM EDT in Pitch f/x Goodies Comment 8 comments

Six rookies got the call and made their major league debuts this week. This will mostly be a run-down of their basic PITCHf/x numbers, but I have a few observations - despite the small samples.

rookie	pitch	#	vs LHH	vs RHH	MPH	PFX_X	PFX_Z	DEG	RPM
Cesar Carrillo	Change	11	6	5	86.2	-7	2	253	1429
Cesar Carrillo	Curve	17	7	10	80.5	7	-4	58	1419
Cesar Carrillo	Sinker	8	8	0	93.9	-8	5	237	1916
Cesar Carrillo	Fastball	26	11	15	94.5	-6	8	215	2063
Esmailin Caridad	Change	1	1	0	81.3	-7	2	252	1278
Esmailin Caridad	Sinker	24	8	16	92.8	-6	5	231	1591
Esmailin Caridad	Fastball	45	26	19	94.4	-4	8	207	1799
Esmailin Caridad	Slider	22	3	19	77.6	7	-2	76	1287
Jeff Manship	Change	5	3	2	81.5	-7	8	224	1964
Jeff Manship	Sinker	12	12	0	90.0	-10	8	231	2499
Jeff Manship	Fastball	12	4	8	90.0	-7	11	212	2620
Jeff Manship	Cutter	2	0	2	84.6	0	5	180	876
Jeff Manship	Slider	5	3	2	80.1	4	0	95	672
Justin Berg	Sinker	13	8	5	92.4	-7	4	240	1781
Justin Berg	Fastball	2	0	2	92.1	-4	5	219	1397
Justin Berg	Slider	12	8	4	83.8	0	4	174	725
Samuel Gervacio	Sinker	5	5	0	92.7	-8	-4	298	1818
Samuel Gervacio	Fastball	9	2	7	95.0	-8	3	249	1801
Samuel Gervacio	Slider	14	2	12	86.4	3	-2	78	804
Trevor Bell	Change	9	8	1	85.8	-3	3	217	856
Trevor Bell	Sinker	1	1	0	91.1	-4	5	221	1181
Trevor Bell	Fastball	47	24	23	93.7	-2	9	191	1935
Trevor Bell	Cutter	5	5	0	85.6	0	7	183	1326
Trevor Bell	Slider	13	3	10	86.2	2	2	135	643

Berg has already been demoted by the Cubs, and his debut actually came during his second cup of coffee. Continuing the Cubs them, Caridad looks like he'll stick around until the next regular comes of the DL and Carrillo will make his first major league start Tuesday when the Padres host Chicago. Carrillo also happens to be a Chicago native who as made his way back from Tommy John surgery.

Since the samples are so very limited, there isn't a whole lot to say. But here are a few notes.

Caridad has managed to get a .31 whiff rate on his fastball, which is impressive
Gervacio has shown an early ability to get grounders, while Carrillo and Caridad have not
More Carrillo - worst strike rate, whiff rate and SLGCON in the Small Sample Derby
Manship threw the most strikes (24/36)
Bell's slider may need to be revisited, some may have been splitters

Yea, that's all I got.

0 recs | Comment 8 comments | Share

Story-email Email Printer Print

Comments

Harry, I think it would be great if you did some work on figuring out the relative importance

And the stabilization rate of each of the stats that you have. IMO, they would be much more interesting (not that they aren’t) if they “meant” something.

SIGN JOHN SMOLTZ

by vivaelpujols on Aug 17, 2009 7:29 PM EDT reply actions 0 recs

Differentiating pitches

Hey Harry,

I’m not sure if you’ve answered this somewhere else — please feel free to point me there if you have. But, why, for example, did you decide to split Carrillo’s fastball up into 2 different pitches? They seem awfully close in terms of speed and movement.

by arzdb on Aug 17, 2009 9:03 PM EDT reply actions 0 recs

Spin axis

That’s the main factor, for me, in splitting two- and four-seam fastballs (AKA sinkers and fastballs). In his case, the split was tough to make.; Sinkers were anything with a spin axis >230. You could move that over to 225 and convert a few from one group to the other.

by Harry Pavlidis on Aug 17, 2009 10:29 PM EDT up reply actions 0 recs

I've been trying to classify without spin

more ideas are good to know! thanks

by arzdb on Aug 17, 2009 11:45 PM EDT up reply actions 0 recs

Be sure to go game-by-game

or park-by-park at least, when splitting fastball types.

by Harry Pavlidis on Aug 18, 2009 9:48 AM EDT up reply actions 0 recs

how extreme are the park factors for pitchfx?

I’ve seen the dramatic differences in release point and sometimes velocity — is it fair to assume that they affect anything pitchfx picks up? does it affect some things more than others?

by arzdb on Aug 18, 2009 10:24 AM EDT up reply actions 0 recs

it varies

The whole things is based on a very sophisticated estimate of 8 characteristics of the ball’s flight when it is 50 ft from the back tip of home plate.
lateral and vertical position (in addition to the set 50 ft value for “y”, although I move everything back to y=55)
acceleration in three directions
velocity in three directions

From there, you/Gameday derive speed, movement and other spin characteristics. Any variance in the 8 factors that are free to vary (x0,z0,ax,ay,az,vx0,vy0,vz0) will impact the “derived” measures we’re most familiar with (mph, pfx_x,pfx_z,rpm,spin axis).

by Harry Pavlidis on Aug 18, 2009 11:08 AM EDT up reply actions 0 recs

makes sense

I just added a correction for park in and it helped tremendously. I didn’t realize how much tighter it made everything. Thanks for the suggestion.

One more question, going back to what I was originally asking: do you use a cluster sorting algorithm (k-means) and then classify the groups or do you just cut off individual pitches?

by arzdb on Aug 18, 2009 3:39 PM EDT reply actions 0 recs