Slicing The Pitch F/X Leader Boards

by Harry Pavlidis on May 15, 2009 3:00 PM EDT Comment 11 comments

Back to some RV100 (average run value compared to average of 100 pitches) leaders, this time by lateral slice.

"Fat" pitches are the middle ten inches of the plate, "Out" and "In" are the next 7 inches each, outside and inside relative to the hitter respectively, which along with the "Fat" slice comprise the 24-inch wide plate that best matches what's actually called in games. "Wide" and "Tight" are off the plate, outside and inside relative to the hitter, respectively.

Here are the baseline RV100 numbers, across all PITCHf/x data going back to 2006. I mean everything, including WBC, Futures Games, Exhibition, Playoffs and Regular Season. Lower numbers are better for pitchers.

Wide   2.22
Out   -2.47
Fat   -2.55
In    -2.15
Tight  2.16

The stats below are from 2009 regular season games, up through May 14. A minimum 71 pitches was needed to qualify for a slice, which made the qualifying group the Tight leaders rather small.

Good Fat
Sean Burnett	-9.0
Jason Jennings	-8.4
Andrew Bailey	-7.4
Ramon Ramirez	-7.0
Jorge De La Rosa	-6.3
Good In
Mark Buehrle	-7.3
Chan Ho Park	-5.6
Scott Kazmir	-5.5
Justin Masterson	-5.0
Jered Weaver	-4.9
Good Out
Ramon Troncoso	-7.1
Brian Tallet	-6.8
Eric Stults	-5.7
Tim Lincecum	-5.6
Mark Buehrle	-5.5
Good Tight
Josh Johnson	-0.9
Wandy Rodriguez	-0.9
Matt Garza	-0.3
Zack Greinke	-0.3
Johan Santana	-0.1
Good Wide
Dan Haren	-0.8
Javier Vazquez	-0.5
Ross Ohlendorf	-0.5
Kiko Calero	-0.4
Jeff Suppan	-0.2

Bad Fat
Wesley Wright	3.8
Zach Miner	3.1
Glendon Rusch	2.7
Luke Gregerson	2.5
Carlos Silva	2.2
Bad In
Anibal Sanchez	3.2
Andy Sonnanstine	2.8
Gavin Floyd	2.4
Jason Marquis	2.3
Micah Owings	1.3
Bad Out
Andy Sonnanstine	2.3
Oliver Perez	2.0
Guillermo Mota	1.3
Justin Masterson	1.0
Cole Hamels	1.0
Bad Tight
Josh Beckett	5.9
Carlos Zambrano	5.6
Braden Looper	5.4
Daniel Cabrera	5.1
Dallas Braden	3.9
Bad Wide
Micah Owings	5.8
Jon Lester	5.5
Jonathan Sanchez	5.4
Yusmeiro Petit	5.0
Duaner Sanchez	4.9

1 recs | Comment 11 comments | Share

Story-email Email Printer Print

Comments

I hate to knock the math again...

You said in the other piece that 6.5 was average, right? So how can a pitcher be any lower than -6.5?

Texas Rangers Prospects: Martin Perez and Juan Grullon - TexasLeaguers.com

by NoNameOnCard on May 15, 2009 6:43 PM EDT reply actions 0 recs

My guess:

Player A is -2 runs in overall RV100 (that’s basically measuring straight ERA if you adjust the scale somehow, I believe). When he throws “Out” (outside corner) he’s -7 RV100. In the other four slices, he’s +1.25 RV100. In other words, maybe he throws to other slices to get behind in the count, then goes to his good slice and gets outs.

Remember that these will be significantly flukey with so little data, so these are true-talent levels, either.

(I still don’t know why 6.5 is average, though…)

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on May 15, 2009 10:49 PM EDT up reply actions 0 recs

that was an error

6.5 was about 6.5 too high, that was the check that made me find and fix the error. The average comes in just below 0 now, which may be a notch low.

by Harry Pavlidis on May 15, 2009 11:15 PM EDT up reply actions 0 recs

So... "run value" is kind of a misnomer, then?

Since zero is neutral, and positive and negatitve numbers represent the degree of effective and ineffective pitches, respectively, maybe “run factor” would make more sense than “run value”.

I know rv100 isn’t your baby, but if it’s measuring from zero and not some theoretical replacement average, “run value” is pretty misleading. Who do we talk to about this…. ?

Texas Rangers Prospects: Martin Perez and Juan Grullon - TexasLeaguers.com

by NoNameOnCard on May 16, 2009 12:00 AM EDT up reply actions 0 recs

That's what JPS says, but doesn't say

gives the total number of runs the pitch would have saved compared to an average pitch of that type

link

by Harry Pavlidis on May 16, 2009 12:53 AM EDT up reply actions 0 recs

If I'm reading the article correctly...

Your method makes sense now, but his doesn’t.

He says that the rv100 assigned to the pitch types is based on the distribution frequency. I take that to mean that a 14% pitch will be thrown 14 times in the theoretical 100-pitch sample. Using his LHP vs LHB table, 14 * -0.0117 != -0.18 and for the fastball, 55 * 0.0018 != 0.02. What did I miss?

How did you calculate your average rv100 values?

Texas Rangers Prospects: Martin Perez and Juan Grullon - TexasLeaguers.com

by NoNameOnCard on May 16, 2009 1:11 AM EDT up reply actions 0 recs

I think he does what I did

which is simple, but he doesn’t make it sound that way. For each pitch, you look-up the run value. I give 0’s to foul-balls on strike two, and 0’s to HBP (for now, cuz that’s wrong),

Once I have the value for each pitch, I simply average them (in this case, across a pitcher/slice) and multiply by 100 to make the number “meaningful” in the context of a 100 pitch outing.

Technically, I have the runvalues stored in a table, the ball/strike count for each pitch, and the outcome. I join the pitch and runvalue data on the count/outcome, so it’s really all in SQL, using the “average” function and grouping by pitcher and slice.

by Harry Pavlidis on May 16, 2009 1:37 AM EDT up reply actions 0 recs

That makes sense to me.

Do you have any insight on his table? His single pitch numbers don’t appear to have any relationship to his rv100 values.

Texas Rangers Prospects: Martin Perez and Juan Grullon - TexasLeaguers.com

by NoNameOnCard on May 16, 2009 3:52 AM EDT up reply actions 0 recs

I'm going to rebuild the table myself

Clearly define my methodologyfor y’all, update the values, get HBP in there. I’m also incorporating GB/FB/LD data (from Gameday) so, next leaderboards (after Monday’s Layers post) will have some other goodies, too. And, you know, screw up some math, too, just to keep everyone alert.

by Harry Pavlidis on May 16, 2009 9:29 AM EDT up reply actions 0 recs