Fun With Numbers
by Jeff Sullivan on Aug 24, 2009 4:20 PM PDT
in The More You Know
42 comments
As you should all know by now, Fangraphs presents certain plate discipline statistics such as first strike percentage and swing rate on balls out of the strike zone. I don't know why it never dawned on me before, but just now I decided to convert these numbers into expected strike rate for pitchers and then compare the result to actual strike rate to see who's getting calls, and who's getting it in the shorts.
The calculation is pretty simple:
Pitches * Zone% = total # of pitches in the zone
Pitches - (pitches * Zone%) = total # of pitches out of the zone
Total # of pitches out of the zone * O-Swing% = total # of swings on pitches out of the zone
Total # of O-swings + Total # of pitches in the zone = Expected # of strikes
Strikes - expStrikes = Difference
Difference / Pitches = Difference%
Here are the leaderboards for pitchers with a minimum of 500 pitches thrown:
| Top 10 | ||
| Rank | Pitcher | Difference |
| 1 | J.D. Martin | 4.6% |
| 2 | Jake Peavy | 3.0% |
| 3 | J.P. Howell | 3.0% |
| 4 | Max Scherzer | 2.8% |
| 5 | Darren O'Day | 2.8% |
| 6 | Brian Moehler | 2.7% |
| 7 | Derek Lowe | 2.6% |
| 8 | Javier Vazquez | 2.5% |
| 9 | Jamie Moyer | 2.4% |
| 10 | Darren Oliver | 2.3% |
In other words, JD Martin's strike rate is 4.6% higher than we'd expect it to be based on the number of pitches he throws in the zone and the number of swings he gets on pitches out of the zone.
| Bottom 10 | ||
| Rank | Pitcher | Difference |
| 1 | Evan Meek | -3.6% |
| 2 | Carlos Villanueva | -3.1% |
| 3 | Anibal Sanchez | -2.9% |
| 4 | Ronald Belisario | -2.7% |
| 5 | Luke French | -2.7% |
| 6 | Jason Motte | -2.7% |
| 7 | Mark DiFelice | -2.7% |
| 8 | J.J. Putz | -2.6% |
| 9 | Roman Colon | -2.5% |
| 10 | Jose Valverde | -2.4% |
Evan Meek, meanwhile, has a strike rate that's 3.6% below where we'd expect it to be. Luke French makes the list as well, ranking 5th-lowest out of 382 pitchers.
(Note that there's a league-wide error of just under 0.4% - there have been 341,240 strikes against 339,264 expected strikes over 547,023 total pitches. This is corrected for.)
We've complained about Doug Fister getting jobbed on several occasions, and indeed, the numbers bear it out - he's registered 192 strikes and 203 expected strikes over three starts and four appearances, for a difference of -3.9%. That's the worst rate in baseball among guys who've thrown 300 pitches.
Anyway, for the curious, here's a link to the player spreadsheet, and here's a link to the team spreadsheet. By this method, the Braves have gotten the most calls, while the Brewers have gotten the least.
4 recs |
42 comments
|
Email
Print
Comments
I knew we were getting screwed!
---
http://www.beyondtheboxscore.com
http://www.rightfieldbleachers.com \\ twitter
by Jack Moore on Aug 24, 2009 4:24 PM PDT reply actions 0 recs
I took a look at something similar to this a couple days ago
http://www.hardballtimes.com/main/blog_article/measuring-the-umpires-affect-on-the-game/
I definitely agree that guys can get unlucky with their strike rates (and subsequently their BB and K rates). I believe that eliminating the umpire affect would take out a lot more variance in even already good stats like tRA.
Thanks
by vivaelpujols on Aug 24, 2009 4:27 PM PDT reply actions 0 recs
Awesome stuff, Jeff.
How much work would be involved in breaking down expected and actual strikes per count and using that against the expected wOBA to get a run value? I’d be curious to see if there was much of a difference, and if a particular pitcher is getting jobbed on strike one or two, as opposed to strike three.
I will smash your face into a jelly.
by Phildopip on Aug 24, 2009 4:55 PM PDT reply actions 0 recs
Dan Turkenkopf found that it was about .161 runs per extra strike
http://blog.stealingfirst.com/2008/04/02/switching-a-ball-to-a-strike/
And their have been several articles that look at the affects of umpires:
http://www.beyondtheboxscore.com/2008/4/5/389840/framing-the-debate
http://www.hardballtimes.com/main/article/a-zone-of-their-own/
And the one that I posted above.
In each case, the affect was found to be very, very high. Dan found that individual catchers can add 25 WINS!!! per season by framing (although a lot of that could be luck). Jon Hale found a similar affect with individual umpires, and I found a similar affect with individual pitchers.
That number is so high, that it is almost certainly wrong; however, their doesn’t appear to be much bias at all or any systematic error (besides maybe accuracy of pitch f/x data, which Mike Fast has found to be somewhat problematic). It appears that umpires just really, really, really suck.
Thanks
by vivaelpujols on Aug 24, 2009 5:02 PM PDT up reply actions 0 recs
*effect
Sorry, that just always bugs me.
by I Lick Squirrels on Aug 24, 2009 5:10 PM PDT up reply actions 0 recs
That's funny, obnoxious grammar Nazism always bugs me.
by acblue on Aug 24, 2009 5:13 PM PDT up reply actions 1 recs
I hate 'alot' a lot.
Just something about how it looks bugs me much more then other grammar mistakes.
The Yankees suck-a-doodle-doo!
by JamMasterJesus on Aug 24, 2009 5:22 PM PDT up reply actions 0 recs
It checks out
I mean, unless pitch f/x data is really inaccurate, I don’t see where a major source of error is. He normalized for batter height, the umpires strike zone, batter handedness…
Thanks
by vivaelpujols on Aug 24, 2009 6:21 PM PDT up reply actions 0 recs
25 wins per season makes the differences between every major league team the catchers, more or less
by Graham on Aug 24, 2009 6:22 PM PDT up reply actions 0 recs
I know, I know
First off, I don’t think it is catcher skill. Mainly just umpire inadequacy. I’m sure there is a skill to framing, but I doubt it’s much more than 1 win.
I really do think that umpires can affect the game a lot, but I never thought it would be 25 runs high. I mean, it doesn’t make sense, but I really can’t see a flaw in his, Jon’s or my method.
Using Jeff’s team numbers, we get that the spread is about 6 wins based on umpires. He didn’t use pitch f/x though, he make a model based off of pitch results. His theoretical spread is a lot lower than the observed spread, which leads me to think that their may be a lot of error in the pitch f/x data.
Thanks
by vivaelpujols on Aug 24, 2009 6:30 PM PDT up reply actions 0 recs
So just so I'm clear, you're talking 25 wins as in 250 runs, right?
Because that number is still making my eyes glaze over
by Graham on Aug 24, 2009 6:35 PM PDT up reply actions 0 recs
It's in the article that I linked to above
In 08, the difference in “extra” called strikes between the leader, Greg Zaun, and the trailer, Gerald Laird, was 13.25 per game. Assuming that the catchers player 120 games, that works out to 1590 extra strikes. Multiply that by the value of an extra strike, .161 runs, and you get roughly 255 runs.
For pitchers I found a similar affect, with a SD of about .25 runs per 100 pitches and a spread of 1.5 runs per 100, and Jon Hale found that same affect for umpires.
It really must be an inherent bias using PItch f/x data, but for the life of me, I can’t think of what it is.
Thanks
by vivaelpujols on Aug 24, 2009 6:58 PM PDT up reply actions 0 recs
Do you mean like "cool" ridiculous
or “you must be screwing up” ridiculous?
Smoltz.
by vivaelpujols on Aug 25, 2009 12:29 AM PDT up reply actions 0 recs
.161 runs per strike?
that seems quite high. Plus, it neglects positional effects (what the count was) as well as combinatorial/multiplicative/additive effects (were there other bad calls against the same batter).
A strike called at 0-0 doesn’t have as much effect as a strike called at 3-2, as the hitter can adjust philosophy to be more optimal given the count. In many situations, a batter has more than 3 strikes to play with anyway – they can foul off balls for a while. What’s the run value of a walk anyway? .250? Less than that? How could the (negative) value of an individual strike (disregarding the count) be any more than 1/3 of the (positive) run value of a walk? Okay, I’ll give you that if a strike is called it means a ball wasn’t called, but still… That’s a profound effect from a single ball or strike.
As to combinatorial effects, I would argue that subsequent bad calls have even less of a run value than the first bad call – the player is already getting shafted and decreasing their run value. What’s a little more shaft?
by PagsBrewCrew on Aug 25, 2009 9:51 AM PDT up reply actions 0 recs
That's where I'd start
The RE swing from a 1-0 count to 0-1 is pretty big, but it’s not close to 0.161. For other counts/situations, my understanding is that it’s much less…. here’s one example.
0.161 seems very high, but I’ll check out the links in this subthread.
by marc w on Aug 25, 2009 10:26 AM PDT up reply actions 0 recs
OK, read the article....
The .161 figure is the result of summing the change in weighted RE/run value in every count. I’m not sure about that as a methodology, but I’ll have to think about it more.
Clearly the effect of turning a ball to a strike (or vice versa) in SOME counts is at least .161, but not most cases.
by marc w on Aug 25, 2009 10:31 AM PDT up reply actions 0 recs
Even if it's half of that
The umpire effect is still ridiculously high.
Smoltz.
by vivaelpujols on Aug 25, 2009 2:37 PM PDT up reply actions 0 recs
This is super useful stuff!
How come no one has thought of this before?
The Yankees suck-a-doodle-doo!
by JamMasterJesus on Aug 24, 2009 5:23 PM PDT reply actions 0 recs
One well-placed baseball allows you to see the world differently.
by abender20 on Aug 24, 2009 5:26 PM PDT up reply actions 1 recs
This should be a semi-mainstream stat
just like the ‘batters luck’ stat (I forget what it’s called)
The Yankees suck-a-doodle-doo!
by JamMasterJesus on Aug 24, 2009 5:37 PM PDT up reply actions 0 recs
No,
I agree with YOU!
The Yankees suck-a-doodle-doo!
by JamMasterJesus on Aug 26, 2009 12:44 AM PDT up reply actions 0 recs
cool.
Glad you did this. During a series a little while back, it seemed like the Yankees were getting a much larger strike zone while the M’s were getting screwed. I wondered at the time if there was a correlation showing umpires favoring big market teams, but figured it’d take way too much work to parse the data.
Interesting that my hypothesis was totally wrong. NYY and BOS (the two AL teams I thought would get the most benefit of the doubt) are middle of the pack.
by Dave Clapper on Aug 24, 2009 5:29 PM PDT reply actions 0 recs
Does this all fall along a normal distribution?
Or are there are things pitchers are doing that either help or hurt them in this regard? Or both?
by greymstreet on Aug 24, 2009 5:37 PM PDT reply actions 0 recs
It's pretty much normally distributed
Thanks
by vivaelpujols on Aug 24, 2009 5:59 PM PDT up reply actions 0 recs
Great stuff
I think it would be interesting to watch rookies vs. crafty veterans and see if there is a big gap in the calls they get. My dad always claimed Greg Maddux never even needed to be close to the plate to get a strike, now we can tell.
by bhsmarine on Aug 24, 2009 5:52 PM PDT reply actions 0 recs
My dad hates Greg Maddux
He came in exactly average over the years 02-08 at -0.005. Oh well…
by bhsmarine on Aug 24, 2009 6:16 PM PDT up reply actions 0 recs
Awesome connect, Jeff.
Next step: look for patterns of who’s getting jobbed and who’s getting extra calls. Maybe breaking balls get missed more often? Maybe older pitchers actually do get more calls? Maybe better pitchers get more calls? Maybe certain stadiums are tougher to call?
Beyond the Boxscore Not a member? Sign up.
by Sky Kalkman on Aug 24, 2009 7:12 PM PDT reply actions 0 recs
Tons of possible correlations to run
I’m excited!
by Jeff Sullivan on Aug 24, 2009 8:17 PM PDT up reply actions 0 recs
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2010 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
