Looking to Refine FIPs with Pitch F/X

by Jeff Zimmerman on Jul 18, 2009 6:00 AM EDT

FIPs measures 3 factors that totally dependent on the pitcher, strikeouts, walks and home runs. A player's strikeout and base on balls numbers are fairly consistent if a player changes ball parks, but the number of home runs he gives up may change depending on the park. The Hardball Times tracks xFIPs which counts every 11^th (league average) fly ball a home run.

A while back I did a piece on Randy Johnson for Driveline Mechanics and found that Randy was leaving the ball higher in the strike zone than he did previously. Most of the home runs he allows was higher in the zone and this led him giving up more home runs this season..

Putting these two ideas together, I looked to see if a pitcher's home run rate correlates to the number of pitches thrown into the heart of the plate (area where the average home runs is hit from) . First I took at all the Pitch F/X available for 2007 to 2009 (I ran the queries a couple of weeks ago, so the 2009 data is from around July 1^st). I took the normal strike zone of 1.5 to 3.5 vertical and -1.2 to 1.2 horizontally and saw how many pitchers where in this zone and how many home runs were hit. I began to shrink the zone down and noting the percentage of home runs hit from that area until I got the the area where 66% and 50% of the home runs was hit from. The following graphic and chart shows the 3 zones:

	Normal Zone	2/3 Home Run Zone	½ Home Run Zone
Percent of HR hit from Zone	94.90%	66.41%	49.85%
X coordinate	-1.2	-1	-0.86
X coordinate	1.2	1	0.86
Y coordinate	1.5	2	2.14
Y coordinate	3.5	3	2.86
Area of Zone	4.8	2	1.24

After creating the three zones, I looked at the 88 qualified pitchers from 2008.and saw how many home runs they gave up when throwing pitches into these zones. I graphed the percentage of pitches (using zone, not umpire calls) in the zone to the number of home runs to see how well the data correlated as seen in the following graphs (one of the 66% HR zone and one of the 50% HR zone):

66_medium

50_medium

There is very little correlation (r-squared of 0.05 and 0.11)

to pitching to the zone where home runs's are given up and actually giving up a home run. A little Master of the Obvious here, but it looks like different pitchers have different abilities. I went on to compare throwing to the heart of the strike zone to the following variables and not one had an r-squared over 0.07: BABIP, GB/FB, LD%, GB%, FB%, HR/FB, HR/9, ERA, FIP.

What I was looking for doesn't seem to simply exist and I have thought of few way to modify search. Since I have the data readily available, here are the 2008 leaders and laggards in Heart Pitches per HR with the average value at 32.1:

Name	Pitches in Heart/HR
Jair Jurrjens	101.7
Jon Lester	96.3
Brandon Webb	86.8
John Danks	86.0
Roy Halladay	80.2
Mike Pelfrey	79.2
Dana Eveland	74.5
Hiroki Kuroda	66.2
Cliff Lee	66.2
Ian Snell	65.5

Todd Wellemeyer	21.1
Paul Byrd	20.7
Tim Wakefield	20.4
Aaron Harang	19.9
Ricky Nolasco	19.7
Vicente Padilla	19.5
Jeff Suppan	19.1
Armando Galarraga	18.6
Scott Olsen	17.5
Brandon Backe	16.4

I didn't find exactly what I was looking for when I began looking at this data. I feel Pitch F/X numbers can be used to help determine the true value of pitchers and batters, but this method does not seem to be one of those cases. I might divide out speed, type and movement of pitch at a later date, but for now I am settled on using FIPs and xFIPs.

Tweet Comment 17 comments | 0 recs |

Do you like this story?

Comments

What about looking at team related HR/FB

I tried this and for now I am calling it PAxFIP. It’s obvious that some pitchers will never regress to the league average (think Matt Cain). I made 4 year running averages for team HR/FB (gotta work on the numbers for teams with new stadiums) and then ran them in a similar fasion to xFIP.

http://www.rotosavants.com/2009/07/paxfip.html
http://www.rotosavants.com/2009/07/trades-ive-made-part-two.html

by TroyPatterson on Jul 18, 2009 8:05 AM EDT reply actions

How about using HR/FB park factors instead of a team's average HR/FB rate.

While your approach could even out, it very well could be biased by a few pitchers with non-average HR/FB rates. At the very least, include visiting pitchers, too. And a park factor would be best.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 18, 2009 10:08 AM EDT up reply actions

Definitely a good idea, though.

And quite helpful for fantasy where you only want to neutralize luck, not park. In fact, I’d build a part of the equation where you plug in BABIP, as a pitcher with a bad defense is less valuable (to fantasy) than a pitcher with a good defense.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 18, 2009 10:09 AM EDT up reply actions

Fantasy

Right. The point of my calculation is mainly for fantasy and while it isn’t truly a park adjustment it does take into account both park and the uneven schedule for league factor. I think BABIP would be a huge addition when you look at a team like the Texas Rangers defense getting better and helping the team.

I think I will look to add that in as well.

by TroyPatterson on Jul 18, 2009 2:54 PM EDT up reply actions

Do you know of a source for HR/FB park factors?

Someone asked me if I had a source for them this week so I’ve been attempting to put them together.

But if one’s already out there, then I don’t need to.

by Dan Turkenkopf on Jul 18, 2009 12:31 PM EDT up reply actions

Ask Brian Cartwright

He has a park factor on everything

Jeff Zimmerman - Protecting the world from RBI's and Wins from my mom's guest house.

by Jeff Zimmerman on Jul 18, 2009 12:32 PM EDT up reply actions

Harry's the king of park factors these days, ask him.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 18, 2009 2:30 PM EDT up reply actions

Statcorner has them

And THT probably does too.

Derosa.

by vivaelpujols on Jul 19, 2009 3:25 AM EDT up reply actions

Thanks

I didn’t think to check Statcorner.

I’ll need to talk to Graham though, because my numbers are way off from theirs.

by Dan Turkenkopf on Jul 19, 2009 11:03 AM EDT up reply actions

Could you also ask him about his Coors field park adjustments?

They seem to be way less than anyone else’s.

Derosa.

by vivaelpujols on Jul 19, 2009 9:34 PM EDT up reply actions

Maybe you've stumbled upon a different finding by accident?

Maybe this is a great way to measure “stuff”. For each pitcher, I wonder how much correlation there is in HR/heart from year to year. Or if we’re just seeing fluke leaders and trailers. Of course “stuff” for preventing home runs could simply be a sinkers instead of four-seamers and consistently lower vs. consistently higher. Could.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 18, 2009 10:04 AM EDT reply actions

Needle in a haystack...

I’ve been trying to figure out, for a while now, some sort of way to evaluate them based on the process of pitching rather than the results. A strike or hit would be a result whereas pitching to a certain location with a certain type of pitch would be looking at the process.

After lots of trial and error, I can’t help but start to think that HRs, to some degree, are just more a product of the hitters than the pitchers. There’s nothing that I’ve looked at to even come close to correlating well with it.

I’d love to keep searching though, hopefully someone can provide some assistance on areas to look at.

by arzdb on Jul 18, 2009 10:55 AM EDT reply actions

I sort of agree and with FIPs becoming more and popular as the standard measure for a pitcher, they should be measured by their HR rate alone

Jeff Zimmerman - Protecting the world from RBI's and Wins from my mom's guest house.

by Jeff Zimmerman on Jul 18, 2009 12:31 PM EDT up reply actions

For my run value stuff

I’m using expected outcomes, park adjusted. I’ll post something on the event/batted ball types in Gameday, maybe tomorrow.

by Harry Pavlidis on Jul 18, 2009 2:46 PM EDT up reply actions

One cliche is that pitchers allow fly balls, but hitters (and parks) determine how many fly balls are home runs.

Like anything, figuring out the extent to which that’s true is the key.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 18, 2009 2:32 PM EDT up reply actions

Completely agree with that senitment

Now, how can we classify a pitch as a hanger, or perhaps, slightly differently, as a pitch that is most easy for a hitter to make solid contact with? How often are these hanging pitches really hit for homers?

I’m interested to see Harry’s post on expected outcomes.

by arzdb on Jul 18, 2009 3:07 PM EDT up reply actions

Along these lines, I'd like to see a study park-adjusting and regressing HR/FB rates.

Then take the high and low group and see what they have in common, probably with pitch f/x data. Do they throw certain pitches? Is it certain pitches/locations that get hit for home runs more often in the air? is it a velocity thing? Or none of the above and it’s more of a “pitchability” issue. That is, do pitchers lower their HR/FB by being smarter and locating better?

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Jul 18, 2009 4:21 PM EDT up reply actions