THE BOOK--Playing The Percentages In Baseball

I little look at Farnsworth I did a few weeks back. 

http://www.royalsreview.com/2009/5/20/881370/quick-look-at-farnsworths

Statement of the obvious here but when Farnsworth throw the ball down the middle of the plate, he gets rocked, but when he works the corners he is fine.

Dan Brooks, as a joke actually, created a measure called Nibbleness that measures how much a pitcher works the edges vice putting it down the middle of the plate.  So far it seems to correlate very well to good and bad outings for pitchers in what little bit I have used so far.

The thing with Farnsworth this year is that he seems to pitch pretty well in low leverage and horrible in high leverage.

What I’m curious about is if he has been this way his whole career or is this just an aberration

Which has led me to wonder that we see tons of studies on “Clutch Hitting” but I feel I’ve only seen one or two on “Clutch Pitching” If people on this board could direct me to more detailed studies on Clutch Pitching (i.e., are there pitchers who pitch great in low-leverage but blow it in high-leverage) that would be great.

Tim: you are walking into the middle of a conversation.  We’ll do our best to ramp you up. 

1. The Book (which you can read on Amazon for free) has a section on “Pitching to the score”.

2. If you click on the Farnsworth link at the top, this will bring you to the Fangraphs page.  Go near the bottom, and you will see “pLI”, which is “Leverage Index”.  More than 1, means higher leverage.  Less than 1, means lower leverage.  His LI this year is 0.55, which means he pitches alot in low-leverage situations.  Go across to his “Clutch” score.  That tells you how well he pitches every year.  As you can see, what he did this year is unprecedented.

3.
http://www.baseball-reference.com/players/split.cgi?n1=farnsky01&year=Career&t=p#lever

Those are his career totals in high (great than 1.5 LI), medium, and low (less than 0.7) leverage situations.  He’s pretty even across the board for his career.

But, if you click on the High LEverage link, you will see his year-by-year breakdown.  He stinks this year: 13 PA, 8 hits (with 1 2b, 2 HR), 1 walk. 

***

There are other excellent articles on the subject.  e.g., Joe Sheehan had a fantastic one on Jack Morris a few years ago.  You can easily find that one.

It should go without saying that 13 PA means virtually nothing (with respect to what it truly tells us about Farnsworth).  I’m sure I can find one 13 PA sequence where even Rivera and Pedro put up a Farnsworth line.

Those 13 PA tells us everything about what actually happened and how fans felt.  We can’t necessarily say that Farnsworth deserves that scorn.

Fangraphs has him being ever-so-slightly unclutch through his career, though most of that has to do with his incredibly unclutchness this year.

http://www.fangraphs.com/statss.aspx?playerid=278&position=P

According to their numbers, for 2005-2008, he was actually slightly to somewhat clutch all 4 years.  (Or should I say, his performance was clutch).

I think the only place Farnsworth may be able to provide value is a place like PetCo (or any other environment that depresses home runs).

The link to the Poz article is not working right now so I cannot comment on it.

Tim, of course lots of pitchers have pitched poorly in high leverage situations and well in low leverage situations, and vice versa, just like lots of batters have hit much better in the clutch or much worse in the clutch during some finite period of time, even over their careers. 

That is not an interesting piece of information (because it is a given) and one that requires no study.  The question you are after is if that kind of “split” performance is by virtue of some kind of significant underlying “clutch” skill or it occurs mostly by chance.  That requires some fairly advanced statistical analysis, some of which has been done for The Book and some of which you can find in other venues.  You are right in that there has been more research in this and similar areas done on hitting than on pitching.  But, you will probably find that there is not a whole lot of evidence that pitchers as a whole have an inherent ability to pitch to the score or in the clutch, as with hitters.  What I DID find in my research for The Book was that while hot and cold streaks appeared to have absolutely no predictive value whatsoever for hitters, that for pitchers, their performance in the last few games seemed to have some significant predictive value (to the tune of .1 or .2 runs per 9 - e.g., a pitcher who has been “hot” will outperform his “regular” projection by .1 or .2 runs in ERA over the next few games or so) over and above their long-term historical performance, i.e., their regular projection based on that historical performance.  Now whether that was due to pitchers being “on” or “off” for finite periods of time, whether it has to do with health issues, or something else, was not real clear to me and probably merits further research and analysis.

Poz link works fine for me…

OK, the link works now.  I don’t see anything wrong with the blog post.  Fairly interesting stories and stats about Farnsworth.

I thought you might have been turned off by Poz not noting his super-high HR rate, as that explains the gap that Poz didn’t see.

But, yes, that aside, a typically fine Poz article.

Re: Clutch score

This info may be at Fangraphs, but I’ll ask here since somebody else might be wondering. What exactly does clutch score measure? Is the number in runs? IOW, does a clutch score of 1.7 mean a player was 1.7 runs better in the clutch? Is the number a percentage? Just curious as to what exactly the number represents.

JD: Clutch = (WPA / pLI) - (WPA/LI)
From FanGraphs:
Get to Know: Clutch
All About Clutch

JD, I think it’s actually wins.

It’s the number of wins that a player produces in that particular game state minus what he would have produce in a neutral game state.