A Thought on Penny

by Harry Pavlidis on Jan 1, 2009 8:15 PM EST Comment 24 comments

The Red Sox seemed to have taken a good risk by signing Brad Penny. When he's healthy, he's an impressive power pitcher. By that, I mean 2008 was not impressive.

One of the things that struck me is the increase in home runs. Since I've been playing around with some new ideas on pitching analysis, I figured I'd run Penny out as a trial balloon.

I've picked 5 metrics to use for a look at Penny's fastball. I'm interested in the trend by month since 2007 using all the available PITCHf/x data on Penny.

ISZ (rate of pitches in the rule-book zone; top/bottom from the average value set by the PFX operators)
FatIns (taking the slices - see the link - and using the Fat and in "inside" "Side" slice to make a 12 inch "hot zone"; rate is like ISZ, but just for the 12 inch area, not the full 17 inch plate)
HR (this rate is simply balls hit for home runs divided by Swings)
Whiff (misses divided by swings)
TBP (total bases per pitch - add up the bases on all the hits on a given pitch, divide by all the times that pitch was thrown - league is around .1)

For reference, I've included the # of fastballs, along with their average velocity, for each month. This is critical, because the September data is limited to two relief appearances. So, the sample is small, but does reflect the horrific performance.

Also note, no data from July 2008 due to DL stint, and not a lot of pitches for March 2008.

In a nutshell, he lost some velocity, missed less bats and got hit harder. Shocker.

Not like anything of note is shown above on it, but the relationship between ISZ and FatIns is of interest to me. If pitchers can throw strikes, but keep the ball either well inside or outside "enough", they'll have more success over time (and I'm not yet considering height). The percentage of strikes in the FatIns area is close to 2/3 from what I've quickly checked so far. Anyway, back to Penny.

What matters for Brad is just the last three lines.

Penny's TBP suddenly hopped above league average in May, and sky-rocketed in the two relief appearances. HR rate when from zero, and/or virtually zero, and started climbing and driving that TBP at the same time.

It's hard to take much from the whiff rates, other than August and September fall below any other month covered. So, I wouldn't exaggerate the meaning of September's garbage, beyond it continuing the trend.

Now that's he's healthy, the question is will he indeed return to pre-injury levels, and for how long. The Red Sox only put a bet down for a single year, and that seems good all around.

0 recs | Comment 24 comments

Story-email Email Printer Print

Comments

Do MLB teams look at this kind of data?

If not, they should seriously consider it. VERY impressive.

Clutch: A measurement of how much better or worse a player does in high leverage situations than he would have done in a context neutral environment. http://www.fangraphs.com/blogs/index.php/glossary/

by bs.uf15bosox9bears23 on Jan 1, 2009 8:28 PM EST reply actions 0 recs

If you mean PitchF/x and the like, yes.

by R.J. Anderson on Jan 1, 2009 8:41 PM EST up reply actions 0 recs

Some do, some don't

There are some folks from MLB that regular read my blog (cubs f/x), but, as of May at the PFX Summit, no team admitted to using it for anything yet. Manny Acta knows about it. Some players know about it. And, don’t forget, MLB teams have video galore these days, which, when combined with PFX, can eliminate the need for some advance scouting trips. I raised that, got a few strange looks, but now, the Mariners are going that route (although they didn’t mention PFX in the articles I saw, just video).

by Harry Pavlidis on Jan 1, 2009 9:43 PM EST up reply actions 0 recs

This is awesome stuff

by Dan Turkenkopf on Jan 1, 2009 10:15 PM EST up reply actions 0 recs

A few orgs are trying to make scouting a science with PitchF/.

by R.J. Anderson on Jan 1, 2009 11:12 PM EST up reply actions 0 recs

They should

MLBAM serves up the data nicely for them. And as long as they don’t even use it to analyze umpires, MLB wants them to use it.

Hey, Dan ….. we need to talk about catcher blocks using PFX to figure out where the ball “should” hit dirt and see if we can quantify blocking skills by situation (count, runner on third, pitch type etc). For example, is blocking an 0-2 slider with a guy on third easier than blocking a 1-0 heater in the dirt with a guy on first. Is it a separate skill?

by Harry Pavlidis on Jan 1, 2009 11:16 PM EST up reply actions 0 recs

This sounds delicious.

by R.J. Anderson on Jan 1, 2009 11:22 PM EST up reply actions 0 recs

We use numbers and stuff. Man.

Webmaster of Driveline Mechanics
http://www.drivelinemechanics.com - An Unconventional Look at Scouting

by Kyle Boddy on Jan 1, 2009 11:58 PM EST up reply actions 0 recs

Definitely

To this point I’ve only been using balls coded as being blocked as opposed to actual PFX coordinates. I feel like when I tried to correlate the two there were a lot of discrepancies. Let me start by looking at how closely the two match.

by Dan Turkenkopf on Jan 2, 2009 10:43 AM EST up reply actions 0 recs

are you tracking it to the ground?

or to the plate?

by Harry Pavlidis on Jan 2, 2009 10:45 AM EST up reply actions 0 recs

Only to the plate

Not enough of a physicist to extrapolate any further than Gameday provides.

Do you have a way to extend that back to see if it hits the ground before it gets to the catcher?

by Dan Turkenkopf on Jan 2, 2009 10:46 AM EST up reply actions 0 recs

Speaking of

How does PFX get negative numbers for pz?

by Dan Turkenkopf on Jan 2, 2009 11:01 AM EST up reply actions 0 recs

not 100% sure

but I believe it is modeled based on the fact that the ball hit the ground at a certain point etc etc. I’ve checked it out. It has Rich Hill’s last MLB fastball in the dirt just like it should.

by Harry Pavlidis on Jan 2, 2009 11:37 AM EST up reply actions 0 recs

I can do that

by Harry Pavlidis on Jan 2, 2009 11:36 AM EST up reply actions 0 recs

not statistical analysis at its best . . .

I am a big fan of this blog, but it pains me when I see a complicated graph and argument condensed into the following conclusion: “Now that’s he’s healthy, the question is will he indeed return to pre-injury levels, and for how long. The Red Sox only put a bet down for a single year, and that seems good all around.”

Is that really the conclusion from all this data? You certainly didn’t need the data to tell you that. You need to present your data in a way that highlights what is new or innovative about it, and avoid falling into baseball common wisdom. That’s what sabermetric analysis is all about.
Another example – you say that the data shows that pitching strikes to the inside and outside is key to a pitcher’s success… well, every broadcaster who can’t tell ISO from OBP knows that controlling the corners is important to a pitcher… you need to explain how the data you collected and analyzed give a statistical context to this commonplace, and how it has the potential to affect how we understand a pitcher’s “control.” Otherwise, it just seems like you’re using a fancy graph to restate received wisdom.
Rant over.
This is one of my favorite baseball blogs, btw – that’s why it bothers me to see this data not used for maximum effect. Keep up the good work.

by scott brosius on Jan 2, 2009 10:03 AM EST reply actions 0 recs

not intended to be

a thought, a trial balloon. A work in progress. Follow along, read some of the earlier posts on pitchf/x. We’re working on new ideas and I’m sharing the sausage making process.

by Harry Pavlidis on Jan 2, 2009 10:12 AM EST up reply actions 0 recs

Which is a great benefit of a blog.

Journal publication is a bit farther down the road.

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

by Sky Kalkman on Jan 5, 2009 6:23 PM EST up reply actions 0 recs

i love sausages.

I guess i just wanted more context – how does it fit into the series on pitch f/x? what preliminary conclusions does this suggest? etc. but sorry for the snarky title of the previous post – i realize it was off the mark, and that this is part of an ongoing thought process…will keep reading.

by scott brosius on Jan 2, 2009 10:17 AM EST reply actions 0 recs

no problem

And, believe me, I appreciate the feedback.

The bigger question, which I think you’ve already alluded to, is how PFX data can be used to quantify different aspects of pitching. Nastiness, Precision, Deception (and that doesn’t get into movement, speed changes, sequencing) are the big three I’m interested in at the moment.

What this leads me to think, as far as using this approach to look at Penny, it is possible that a slight loss in precision (more fat pitches) can be a large factor in a big bump in TBP (all else being equal). Possibly (all was not equal with this case).

To get at that, the discrete zone(s) needs to be exchanged for something more continuous/probabilistic, plus broader analysis of a larger sample of pitch(er)s and some more focused case studies.

by Harry Pavlidis on Jan 2, 2009 10:44 AM EST up reply actions 0 recs