Testing an Old Adage...Again

Mike Fast has a great piece up over on The Hardball Times researching the correlation between working quickly and effectiveness. While I study I did on Baseball Prospectus last May used average game time and was more historical in that it went back to 1970, Mike uses the time stamps that MLBAM is providing in its Pitchf/x data for 2008.

What Mike found largely corresponds to what I concluded, namely that there doesn't appear to be any relationship between defensive support as measured by defensive efficiency (DER) and BABIP and time between pitches for team or individual pitchers measured relative to their teams (although I was using unearned runs instead of DER as I should have).

He does find, however, that when looking at BABIP in terms of the number of seconds that elapsed since the previous pitch, the BABIP is lower for pitches thrown within 10 seconds and higher for pitches thrown in excess of 50 seconds since the previous pitch (he does throw out pitches that came in a minute or more after the previous pitch). As Mike notes, there are other factors to control for, not the least of which are hit type (line drive, fly ball, ground ball, popup) and pitcher quality and hitter quality. Still, it's pretty interesting stuff and just one of the many applications of Pitchf/x data.

Infinite Pitcher Abuse Points

I'm sure many of you have seen this article related to Japanese high schoolbaseball but several folks sent it my way today and I hadn't...

School team hit for 66 runs in two innings

I especially like this quote:

The hapless hurler had already sent down over 250 pitches, allowing 26 runs in the first inning and 40 in the second before Kawamoto asked for mercy.

"At that pace the pitcher would have thrown around 500 pitches in four innings," Kawamoto's coach was quoted as saying. "There was a danger he could get injured."

Santana and the Changeup

After a little exchange Ken Davidoff at Newsday wrote a little about Johan Santana and his reliance on his changeup in his column on April 6th. The relevant section reads like this:

According to Fox, MLB.com charted 11 of Santana's outings last year, including his relief effort in the All-Star Game. Of 1,033 pitches, Santana threw 61 percent fastballs, 27 percent changeups and 12 percent sliders, which comes close to Bill James' full-season tally (58-29-11). Not surprisingly, Santana used the changeup far more against righty hitters (he threw it 33 percent of the time) than lefties (7 percent).

Just as Morris suggested, Santana does love using the pitch for strikeouts. Of 269 situations that Fox charted in which Santana had a hitter at 0-and-2, 1-and-2 or 2-and-2, he threw the changeup 123 times. Of the 86 strikeouts Fox witnessed, the changeup produced 53 of them.

Interestingly, of the 11 home runs Santana surrendered on non-full, two-strike counts on Fox's watch, just two came on the changeup, with eight from fastballs and one off a slider.

It turns out that Mike Fast did a nice analysis of Santana back in January and as you would imagine found essentially the same thing albeit in much more detail. From a start by start basis the mix of Santana pitches in 10 of his starts and his All-Star appearance last season can be seen below.



From this it is not apparent that he increasingly used his changeup as the season wore on and in fact it shows a trend where he used his fastball a bit more as the season progressed.

Facing Clemens

My column this week on Baseball Prospectus is an interview with the author of the new book Facing Clemens: Hitters on Confronting Baseball's Most Intimidating Pitcher, Jonathan Mayo. Some readers I'm sure will recognize the name since Mayo is also a senior writer for MLB.com who typically writes about the minor leagues and can be found all over the place as the draft approaches.

In epitome, Mayo's book is a look at what it's like to compete against Clemens from the perspective of thirteen hitters who faced him at various points in his and their careers. Beginning with Dave Magadan, whose University of Alabama Crimson Tide faced off against Clemens' University of Texas squad in the 1983 College World Series, Mayo takes us all the way through the 2007 season with Torii Hunter's final three--ultimately unproductive--plate appearances that capped his unbroken string of futility. Along the way we hear from Hall of Famers like Cal Ripken Jr. and future member Ken Griffey Jr., star players including Gary Carter, Chipper Jones, and Luis Gonzales, to lesser-known hitters like Daryl Hamilton and Phil Bradley (Clemens' 20th strikeout victim in his 1986 record-setting performance), and finally culminating with the story of minor leaguer Johnny Drennan who homered off of Clemens during the pitcher's minor league stint in 2006 as well as Clemens' son Koby. In all, thirteen players are interviewed and the book includes plenty of interesting anecdotes not only about Clemens but on other topics from the hitters perspective.

Although readers will no doubt read the book in a different light than Mayo had intended, it is an interesting compilation and it was nice of Jonathan to "sit down" with me for the interview.

Yin and Yang

I thought I'd finish off January with a couple of links...

Lovin' on Bannister. MLB Trade Rumors did a great interview with Royals pitcher Brian Bannister in three parts. Part 3 is where it gets really good as Brian reveals that he does his own statistical analysis (we already knew he was a BP reader) and gives us his take on DIPs theory. Several well-renowned analysts have already started the discussion into his insights and I'm sure we'll be seeing more in the future. What really encourages me about this is recognizing the value that thoughtful players like Bannister of Jeff Francis can provide and it makes me wonder how teams are utilizing those resources in parterning with the analytical resources they have.

Not so Much Insight. Kind of the anti-Bannister kind of observations were offerred by MLB.com reporter Marty Noble back in early January. In a previous article Noble used RBIs per 100 at bats to make a comparison between newly acquired catcher Brian Schneider and departed backstop Paul Lo Duca. In the response I've linked he tries to explain himself and although his main point that Schneider and Lo Duca are no longer as different offensively as some people claim is valid, there's simply no way he can get out of the hole he's dug. He has the right idea, namely that opportunities are important and rate statistics rather than counting stats are key, but of course he fails to select the right kind of opportunities to make the kinds of comparisons he's going for.

That said, he dropped two little gems that I couldn't pass up:

Computers have contributed to a current glut of statistics that, to a degree, distort the picture. We have so many now that we lose focus on what is most important. The objective of the game is to win, and to win a team must outscore its opponent. Nothing, therefore, is more important than runs -- both producing and preventing them.

To what degree and to which statistics is he referring? Actually, I would argue that by translating traditional statistics into the currency of runs assuming an accurrate weighting, the vast majority of the supposed "glut" of statistics (VORP, BaseRuns, Linear Weights, defensive metrics, base running, etc.) have served to paint a more accurrate picture of "what is most important" - creating run differential that leads to winning games.

That Lo Duca might have had a higher on-base percentage or slugging percentage means less to me than the number of runs he produced. The next time a team wins a game because it produced a higher on-base mark and scored fewer runs than its opponent, please alert me.

Here I think there are two points of confusion.

First, it turns out that the very combination of metrics he mentions, on-base percentage and slugging percentage (OPS), is a very strong predictor of runs produced since it accounts for the key ingredients (getting on base, moving runners, and avoiding outs) that are so problematic in looking at things like RBIs per 100 at bats which only measure one part of the equation. Additionally, by not accounting for context nor understanding how other metrics predict offensive output Noble ends up inverting the relationship between offensive production between the statistics he discusses.

Second, in his last sentence he stumbles across the problem of scale. It is tautological to say that run differential is a perfect predictor of wins and losses at the level of an individual game. Therefore RBIs and run scored (at least for the offense) take on primary significance in that context and at that scale while OBP and SLUG are less predictive. However, once you raise the aggregation level, those counting stats take on less significance in player evaluation because a particular player's role in generating offense is about more than the tallying of the end result (an RBI or run scored) to the point where it quickly becomes the case (and well before the level of seasons) that OBP+SLUG and other derivative metrics are more indicative of offensive contribution and therefore wins and losses.

This confusion of effects at various scales reminds me (not coincidentally because I'm now reading this book) of one of the primary themes in the writing of the late Stephen Jay Gould. He often railed against the position of ultra selectionists or adaptationists who insisted that natural selection was the exclusive driver and shaper of the pattern of life on earth. Gould contended that evolution operated differently at different levels through various mechanisms and that what worked at one level did not necessarily have power at another. For example, he argues that while natural selection works through differential reproductive success to build adaptations at the level of individual organisms (coloring, wings, claws, size, etc.) those adaptations may have little or nothing to do with survival at the higher level of species. In one of his favorite examples he liked to point out that the small size and adaptability of mammals during the age of the dinosaurs was likely the result of the domination by dinosaurs in the niches available to larger animals. However, when the meteor struck it was those "negative" traits that allowed the mammals to survive but doomed the dinosaurs.

Profiling a Ray

The always fascingating Marc Normandin allowed me to horn in on his regular gig this week and contribute a little PITCHf/x analysis to his player profile of the Rays (not Devil) James Shields. Unfortunately Shields had just eight of his 31 starts recorded by PITCHf/x but still the 712 pitches does allow us to see some definite patterns. The article does not require a subscription.

PITCHf/x Musings

Many of the colunms I wrote discussing the PITCHf/x data made available through MLB.com's Gameday system are now available sans subscription on Baseball Prospectus. Those articles are:

October 25, 2007.
Schrodinger's Bat: Free Stuff and the Men in Blue.
Postseason umpiring and an early holiday present for our readers.

October 11, 2007.
Schrodinger's Bat: On Atmosphere, Probability, and Prediction.
Ranging across a couple of old and new themes, explaining that there's something about the weather, and Pythagoras can rock steady.

August 23, 2007.
Schrodinger's Bat: Visualizing Pitches.
After digging through this data, you'll no longer wonder why they say hitting is the hardest thing to do in sports.

August 16, 2007.
Schrodinger's Bat: Putting the Pedal to the Metal.
What happens when pitching in a pinch? Do pitchers have something extra that they can put on the ball when they're in a jam?

July 26, 2007.
Schrodinger's Bat: Calling the Balls and Strikes.
A look umpire tendencies to see how much human error plays a role in calling pitches.

July 5, 2007.
Schrodinger's Bat: Searching for the Gyroball.
Is it there, or isn't it? Dan dives into Dice-K's data to find out.

June 28, 2007.
Schrodinger's Bat: Playing Favorites.
Parsing the data can help us address questions of bias among umpires in calling balls and strikes.

June 21, 2007.
Schrodinger's Bat: Gameday Meets the Knuckleball.
Dan continues his series using pitch data by examining the case of Tim Wakefield.

June 14, 2007.
Schrodinger's Bat: The Science and Art of Building a Better Pitcher Profile.
Popping the hood on King Felix as a demonstration of what's possible with PITCHf/x data

June 7, 2007.
Schrodinger's Bat: Gameday Triple Play.
How different ballparks affect velocity, whether pitchers use the fastball more early in games, and the challenge of quantifying plate discipline.

May 31, 2007.
Schrodinger's Bat: Physics on Display.
Further adventures in pitch-by-pitch data.

May 24, 2007.
Schrodinger's Bat: Batter Versus Pitcher, Gameday Style.
Evaluating the strike zone, the umpires, and some large-scale issues with a tremendous new tool.

May 10, 2007.
Schrodinger's Bat: Phil Hughes, Pitch by Pitch.
Dan uses MLBAM data to reconstruct the no-hitter that wasn't.

Gameday Video

There is a very nice video that explains a bit about the PITCHf/x system over on the Gameday blog. In other PITCHf/x news Joe P. Sheehan has another nice article on sinkers over at Baseball Analysts and tomorrow my column will take another look at plate discipline. And of course anyone interested in this topic should be keeping up with Mike Fast and the work he's doing over on his blog. In particualr he beats me to the punch and uses the approximations given by Dr. Nathan to start calculating spin direction and spin rate. Very cool.

Jimenez Looking Good

Tonight Rockies rookie Ubaldo Jimenez turned in a good start for the third consecutive outing beating the Nationals here at Coors Field. I chronicled his arsenal over on the Rocky Mountain SABR site using PITCHf/x data.

Update: Mike Fast and Sky Kalkman point out that the data used to plot the fastball was incorrect. I inadvertantly used a positive rather than a negative vertical acceleration which caused the pitch to appear to level out. I've since corrected the graphs in the article at RMSABR. My apologies.

Tip of the Iceberg

A few more links related to research with the new Gameday data.

It’s a Pitch-by-Pitch Scouting Report, Minus the Scout. This article by Dan Rosenheck appeared in the Keeping Score column in the New York Times over the weekend. He references a few of the columns I've written at BP in the following comments:

Most studies have focused on classifying the characteristics of various pitches — Félix Hernández’s four-seam fastball is usually thrown between 94 and 97 miles an hour and breaks around 8 inches toward a right-handed batter — and using them to generate profiles of pitchers (he only throws his changeup 3 percent of the time versus right-handed hitters).

Some work has also been done on identifying batters’ tendencies: Iván Rodríguez swings at nearly 60 percent of pitches thrown to him out of the strike zone, and Juan Pierre makes contact with 92 percent of the balls out of the zone he swings at, for example.

And in talking with Dan as he prepared the piece we discussed the fact that this data provides quantification to concepts that are already well understood in terms of advanced scouting. As Dan says:

“Will chase curveballs low and away” will become “swung and missed at 73 percent of pitches thrown under 83 m.p.h. with a vertical break of at least 12 inches on two-strike counts on the outer third of the plate.”

“Slider lacks bite” could be replaced by “slider begins to break 30 feet from home plate.”

However, it should be noted that pitches aside from the knuckleball do not have early or late break as implied by his comments on sliders and instead break in a uniform way as they travel from the pitcher's hand to home plate.

Two of the aspects that we discussed that I think are particularly interesting he described this way.

The data could be used to evaluate prospects, by answering questions like, “Will he ever learn to lay off a breaking ball?” or to better understand park effects, by revealing just how much movement a particular pitcher could expect to lose from his slider at Coors Field.

By quantifying the characteristics of pitches and building up a historical record we'll be able to ask questions related to age and development across pitch profiles (velocity, trajectory, location, and spin). So for example, it may turn out that certain types of hitters have trouble with certain pitch profiles but that they tend to learn to recognize and lay off the pitch or put it into play with greater success as they age or gain experience. There may be other types of hitters for which this is not true and having the data will at least allow us to ask the question. Of course with historical data the mirror questions can be asked of pitchers as well.

In addition I think we're learning that there are discernible differences in how pitches behave under the different conditions in various parks. PETCO Park for example with its heavier sea air both causes pitches to decelerate more and allows for greater break on spinning pitches. Understanding just what those affects are may allow us to create "pitch profile park effects" that more accurately enable us to predict how a pitcher might fare in a different environment. I've written a bit on this subject already and have been working some with Alan Nathan, a physicist and head of SABR's Science of Baseball committee from the University of Illinois, on this very question and should have some things to share in the near future.

Finally, Dan goes on to say:

But the recent findings represent a tiny fraction of the research that the data will ultimately make possible. Eventually, a large portion of the tasks now done by major league scouts — visually evaluating strengths, weaknesses and trends — will be measured numerically.

While I agree that at the present time we're touching the tip of the proverbial iceberg, I would simply caution that the ability of researchers to ask these questions hinges on two very important conditions. First, as Dan says the data needs to continue to be made available in some form be it subscription based or free. And second, researchers need to understand the limitations of the system not only in terms of accuracy but also variance between ballparks and how the system is being tweaked to provide more accurate data. For example, the in ital point at which pitches are tracked was changed in early June from 55 feet and then experimented with for the rest of the month, settled at 50 feet in early July, and now fluctuating once again in an effort to increase accuracy.

And while I also agree that there are many aspects here that will be quantified and overlap with traditional scouting, it will always be the case that these tools compliment and do not in any sense replace what scouts do. Not only will systems like this not be available in the amateur and minor league circuits for quite some time (not to mention bullpens as Dan mentions), they will be used to augment understanding already gained from traditional methods. For example, in terms of its relationship with bio mechanics analysis like that done by Will Carroll, this system starts after the release point and therefore after everything from tempo to leg kick to balance to arm slot have already taken place.

Under Pressure. Joe P. Sheehan at Baseball Analysts looks at the relation of pitch types to Leverage - something that had not occurred to me. While it's certainly interesting and he shows, for example, that Jake Peavy relies more on his slider than his fastball in pressure situations, I think you'd also have to normalize the data for the base/out and handedness of the batter. It could be that Peavy relies more on his slider in pressure situations because he relies more on it with runners on base which also happen to have higher Leverage indexes.

Strike Zone: Fact vs. Fiction. John Walsh totally steals my thunder by examining the actual dimensions of the strike zone as it is called by major league umpires. What I find interesting is that he notes that right-handed hitters end up having to defend a strike zone that is slightly larger while I've found that left-handers are getting 10% more strikes called against them on pitches out of the strike zone. In looking at John's data I think the reason for this is that left-handers have to defend more territory on the outside part of the plate and pitchers concentrate on this area throwing a disproportionate number of their pitches in that region.

Another look at the sinker. Louis Chao at THT looks at contact rates by pitch types and finds, a little surprisingly, that sinkers have higher contact rates than fastballs. My take is that sinkers drop more in accordance with what the hitter is expecting and so they're able to put the bat on the ball albeit typically driving it into the ground. Four-seam fastballs, on the other hand, do not drop as much as would be expected and so batters swing under them. This is supported by the fact that a four-seamer typically drops 10-15 inches less than the theoretical reference pitch while a sinker drops only 2 to 7 inches less.

Rain Delay Musings

I'm scoring the Rockies/Phillies game at Coors Field tonight and in the very first inning the rains came causing the game to be delayed. So more to entertain myself than you here are a few random thoughts on the passing scene (to borrow a phrase).

Angering the Gods? - Before the rain delay Jimmy Rollins homered to right field on the 3rd pitch of the game from Rodrigo Lopez. After a Chase Utley double off the right field wall with one out Ryan Howard lofted a fly ball down the left field line and it just made its way over the wall for a two-run homer. Before the homer lightning had been evident out beyond left field and I think when Howard hit that ball there was an accompanying roar of thunder. Very strange and more than a little distracting to say the least. Rollins homerun was the 100th of his career and on 6/27 Howard hit his 100th and in the process becoming the fastest player to do so (325 games).

More Pitchf/x - A couple more articles on using the Gameday data for analysis were written by Joe P. Sheehan and John Beamer. In Joe's article I love the idea of looking at this data from a consistency standpoint and am not surprised that Joe doesn't see much of a difference between starts. While it's probably the case that there are some pitchers whose success hinges on good stuff on a particular night, those are likely the more marginal pitchers. And my assumption is that location and pitch selection in specific situations would vary more than a measurement of "stuff". Better pitchers like Halladay simply have good stuff and differences between good and bad starts also have a lot to do with simply luck in which balls are hit at fielders and which aren't. This would be a topic to revisit after we have a few thousand pitches for starters in good and bad starts. John, on the other hand does a profile of Tim Hudson much like I've done for King Felix, Tim Wakefield, and Daisuke Matsuzaka. What he adds, and what I'm just getting around to, is sorting out pitch types using a clustering algorithm. I've recently gotten ahold of an implementation of one such algorithm in my programming language of choice so I'm excited to see how it can be modified or customized for this type of data.

And a Clarification - Related to PITCHf/x one of the questions I get most frequently and haven't explained very well is just what the measurements of vertical and horizontal movement are relative to. In short, they are measured relative to a theoretical "reference pitch" defined as a pitch thrown at the minimu with the same release velocity and release point but with no spin. Therefore a pitch that has a vertical movement of 11 inches like many four-seam fastballs doesn't actually rise four inches but rather drops 11 inches less than the reference pitch. As a result the way to get a more intuitive measurements of the movement of a pitch, especially vertically, one needs to have a feel for the trajectory of the reference pitch. While this can be inferred from looking at the data I don't possess the parameters for calculating exactly the trajectory of the reference pitch.

MLB2K7 Gets a Thumbs Up - I bought this game for XBox 360 a month or two ago and have great fun playing it. Compared to MLB2K6 this is head and shoulders a better game. They've adjusted the way fielding works by allowing you to have your fielders sprint and their reaction times to your movements are much snappier. The graphics are better and there seem to be more signature stances and windups for various players. I've had no issues with the game seeming to go crazy in terms of offense or pitching when playing in franchise mode like I did last year and the various game levels seem to incrementally make the game harder in an appropriate way. I'm playing the Cubs in franchise mode and am hovering at .500 by playing some of the games and managing others. Unfortunately the manager interface has not changed and so it still lacks all of the strategic options of the real-time game play and most annoyingly does not let you warm up pitchers when at bat. One cool thing is that through XBox Live I was able to download Olympic Stadium and they say there will be other retro parks added in the future including the Polo Grounds and Fores Field among others.

Ted and Ted - I didn't realize that Cubs pitcher Ted Lilly was named after Teddy Roosevelt who once employed Lilly's grandfather. Very cool.

Rox Surge? - The Rockies are now 25-16 since May 22nd (second only to Seattle at 28-15) after last night's walk-off win in extras after a Brad Hawpe two-out, homer to dead center in the bottom of the ninth tied the game. The Rox are now 25-18 at home and 18-25 on the road following that miserable 1-9 road trip where they lost in walk-off fashion four times and that saw Brian Fuentes blow four saves. The offense seems to be firing on all cylinders and it's clear that the addition of Ryan Spilborghs after the John Mabry/Steve Finley experiments failed miserably. Spilborghs has 24 RBI with his 26 hits and is 4 for 5 in his most recent pinch-hitting chances and 7 for 17 overall. One of the things that has most impressed me this season in watching the Rockies is how Willy Taveras bunts. He now has 27 bunt hits on the season with the next closest player having 8. He also has 40 infield hits which leads the majors. I haven't looked it up but my guess is that he's successful around 80% of the time when attempting to bunt for a hit which seems pretty remarkable.

The tarp is coming off the field and so we'll be resuming here at some point. Right now, however, the grounds crew is wrestling with the tarp as it catches 20 mph gusts of wind and drags them to and fro.

The One About the Gyroball

In an ongoing series on digging into the PITCHf/x data provided by MLBAM's Gameday sysem, this week in my Schrodinger's Bat column I take a look at Daisuke Matsuzaka. The system has captured 586 of his pitches over six starts and so that gives us a pretty good sample to work with as shown in the following table:

Start              Pitches   IP  H  SO  BB ER
4/17 at Toronto        105  6.0  3  10   3  2
5/9 at Toronto         102  7.0  5   8   3  1
5/25 at Texas           85  5.0  7   6   3  5
6/5 at Oakland          55  7.0  7   8   2  2
6/22 at San Diego      126  6.0  5   9   5  1
6/27 at Seattle        113  8.0  3   8   1  1

Interpreting the data to identify his varied repetoire is somewhat more difficult than it is with pitchers who throw "only" three or four pitches but I was encouraged that it could still be done with some degree of accuracy (as checked against the Inside Edge data made available on ESPN Insider). In the article I identify his fastball, slider, cutter, curve, changeup, and forkball/splitter. His hard sinker, or shuuto, apparently behaves much like the forkball and so some of these may be included in the forkball category.

So what about the gyroball? Well, you'll have to read the article to see if the pitch remains a mystery.

Sinkers

As many of you know I've been writing about the PITCHf/x data captured by the new Gameday system the last several weeks in my Schrodinger's Bat column over on Baseball Prospectus. In answering a question for a colleague I ran a query to take a look at which pitchers have the most sink on their sinking fastball and so I'll share the results here.

There is certainly some difficulty in separating sinking fastballs from four-seamers (in some research on Chad Gaudin I found I couldn't reasonably classify some 5% of his fastballs) since the data is continuous and doesn't come nicely labeled. So as a first approximation I thought I'd take a look at all pitches thrown between 87 and 93 miles per hour and that had the appropriate horizontal break for a fastball in order to weed out any sliders. This is similar to what John Walsh did in an excellent article at THT and builds on the work that Joe P. Sheehan did over at Baseball Analysts. The result is the following table of the top 30 pitchers (pitchers who throw from the side excepted since their vertical movement is actually negative in many cases as John discussed).

Name               Throws  Pitches   AvgVel    Vert   Horiz  MaxVel
Felix Hernandez    R             69    89.7     2.2    -3.5    92.9
Kameron Loe        R            529    89.6     3.7    -7.7    93.0
Derek Lowe         R            575    90.2     3.8   -10.7    93.0
Roy Halladay       R            481    90.7     3.8    -7.5    93.0
Brandon Webb       R            111    89.7     3.9    -9.4    92.9
Julian Tavarez     R            296    90.6     3.9   -10.2    92.9
Aaron Cook         R             82    91.0     4.2    -7.2    93.0
Tim Hudson         R            465    90.8     4.5    -6.8    93.0
Jamey Wright       R             72    89.5     4.7    -8.0    93.0
Jeff Weaver        R            202    89.1     5.5   -10.8    92.8
Scott Downs        L            128    89.3     5.6    11.0    92.2
Jose Contreras     R            321    90.2     6.0    -7.7    93.0
Sergio Mitre       R            107    90.0     6.0    -9.2    92.6
Chad Paronto       R            142    90.0     6.1    -5.8    92.8
Jimmy Speigner     R             61    89.5     6.3    -6.0    92.6
Brad Thompson      R             56    90.0     6.5   -10.2    92.0
Miguel Batista     R            319    91.1     6.5    -6.7    93.0
Paul Maholm        L             50    88.5     6.6     6.5    90.6
Zach Duke          L             55    88.9     6.7    10.0    91.4
Gil Meche          R             60    91.2     6.8    -4.9    93.0
J.J. Putz          R             53    89.7     7.0    -6.2    93.0
Oscar Villarreal   R             93    90.0     7.0    -6.8    92.9
Chad Gaudin        R            437    90.6     7.1    -6.8    93.0
Carlos Zambrano    R            113    90.6     7.1    -5.8    93.0
Sean White         R            175    91.1     7.2    -8.2    92.9
Eric O'Flaherty    L            120    90.2     7.2     6.3    92.8
Jesse Litsch       R             62    89.1     7.2    -5.1    92.8
Kip Wells          R             82    90.6     7.3    -7.1    93.0
Vicente Padilla    R            397    90.9     7.4    -6.9    93.0
Robert Janssen     R            102    90.8     7.4    -3.3    93.0

You'll notice that the vertical movement column is still positive for all these pitchers. That's the case because the value is calculated relative to the movement of a theoretical reference pitch that is spinless but thrown in the same way as the pitch in question.

So then to get a feel for what these vertical measurements mean, we can compare them to some pitchers who do not throw a sinking fastball but who do throw their fastballs in the same velocity range. For example, Brad Penny has thrown 230 pitches in this velocity range with an average vertical movement of 12.1 inches. Brandon McCarthy has thrown 264 with a value of 12.1, Randy Wolf has thrown 456 at 11.1, and John Garland has 585 at 10.7. What this indicates is that a four-seamer thrown in the same range drops 10 to 12 inches less than the theoretical reference pitch and so our sinkerballers throw pitches that sink 6 to 9 inches more than that. This seems realistic and of course the list of pitchers near the top (Hernandez, Lowe, Halladay, Webb, Cook) are all the usual suspects.

It's also interesting to note which pitchers have more tail on their sinkers (a negative horizontal movement indicates tailing into a right-handed hitter). Derek Lowe, with his combination of sink and movement, makes it very difficult on opposing hitters.

The Dance of the Knuckler

This week in my column on Baseball Prospectus after rehashing some history of the pitch thanks to Rob Neyer and Peter Morris, I take a look at the knuckleball of Tim Wakefield from the perspective of PITCHf/x and Gameday. The system has tracked his starts from April 6th at Texas, April 18th and May 10th at Toronto, May 26th back in Arlington, and June 6th at Oakland, a sample which includes 461 pitches.

As you might expect a plot of the horizontal and vertical breaks aren't very informative since the pitch is all over the place and therefore doesn't form the tighter clusters you can see with more traditional pitchers. Still, his three pitches (the fastball and curve) are able to be detected when you add velocity to the mix as his knuckers are very consistently thrown with a release velocity between 65 and 70 miles per hour. His fastballs are in the mid 70s and his curve in the low 60s. After discussing classification the article goes on to answer the questions when he throws his various pitches, where he throws them, and what the results are.

Legos to the Rescue

A couple of interesting tidbits this morning.

A co-worker sent me this link from the local Colorado Springs paper. Teenager Cameron Kruse, who works as a Sky Sox ball boy, built a machine to rub down baseballs in a uniform way. There's also a nice video on YouTube and although it doesn't mention it in the article, he might have used Lego Mindstorms for the programming.

Clay Davenport looks a little deeper into the number of minor league innings rookie pitchers have logged over time. The analysis is still incomplete but is suggestive that perhaps the number of innings accumulated in the minor leagues hasn't declined as has recently been discussed.

The Physics of Drag

My column today on Baseball Prospectus delves once again into the PITCHf/x data tracked by the new Gameday application. This time I take a look at the drag on a pitched ball and square the data with the description of the model discussed by Robert Adair in The Physics of Baseball.

To answer the most frequently asked question thus far - no, I haven't looked at Tim Wakefield in any depth. I did see, however, that his average pitch (and I have 346 to look at) lost exactly 10% of its velocity. Overall though that percentage decrease is in line with the following chart (a version of this chart is also in the original article) since his average pFX (which is a measure of the break of the pitch) was 8.6 and his average start speed was 68.5 miles per hour as shown in the chart below.



However, that percentage does not seem to differ by the break length (a different measure of break introduced this year) nor the pFX value. It's also interesting to note that all but one pitch came out of his hand at less than 79 miles per hour. I think it's likely that the Magnus force placed on a knuckler as it moves in various directions tends to slow it down more than one would other think based on the slow speed and lack of spin.

Deep Data Dive

As promised yesterday my column on Baseball Prospectus this morning dives deeper into the PITCHf/x data tracked by the 2007 version of Gameday (a new update was released on May 10th and is much more performant).

In this article I take a look at the velocity and location data that includes over 40,000 pitches and discover that given a one-inch margin of error the system agrees with umpires to the tune of 90%. Not bad and very similar to the QuesTec results published by Robert Adair in an article titled "Cameras and Computers, or Umpires?" that was published in Volume 32 of SABR's The Baseball Research Journal.

Mad Dog

Greg Maddux is off to a nice start in 2007 with a 3-2 record and 3.20 ERA in 50 2/3 innings having given up only 45 hits and 7 walks. It turns out that of his eight starts this season five have been recorded more or less completely by the Enhanced GameDay system. In all, that's 358 pitches, 164 to right-handers and 194 to left-handers. Just fiddling around with some of the data today here are some random observations.

The fastest pitch leaving his hand was 94 mph. He threw the pitch to Garrett Atkins on April 6th in the top of the third inning. The pitch was called a ball. That pitch was also the fastest when it reached the plate at 83 mph.

The slowest pitch he threw was 71.5 mph to Kazuo Matsui in the 6th inning of that April 6th start. That pitch crossed the plate at 62.6 mph before Matsui hit a ground ball to third for an infield single.

His average velocity out of the hand was 85 mph and crossing the plate was 75.2 mph. But what's most amazing is that the standard deviation of his muzzle velocity was just 2.86 mph. By comparison, of pitchers who have thrown 100 or more pitches in 2007 with GameDay watching his pitches have varied the least. Jason Schmidt was close at 2.95 mph and on the other end Randy Wolf was at 8.85 mph.

The breakdown of the outcomes of his pitches were:

Ball                         102   28.5%
Called Strike                 79   22.1%
In play, out(s)               61   17.0%
Foul                          50   14.0%
Swinging Strike               24    6.7%
In play, no out               19    5.3%
In play, run(s)                6    1.7%
Foul Bunt                      5    1.4%
Pitchout                       4    1.1%
Foul (Runner Going)            3    0.8%
Ball In Dirt                   2    0.6%
Swinging Strike (Blocked       2    0.6%
Missed Bunt                    1    0.3%
                             358

Not Surprisingly he doesn't get many swinging strikes.

The break down of his pitches against lefties and righties is shown in the two graphs below (pictured from the perspective of the pitcher).


Against lefties he clearly stays on the outer half and besides how few pitches he leaves in the middle of the plate, it would appear he gets a pretty good number of calls on balls that are actually outside the strike zone. Cory Schwartz was kind enough to answer a few questions on the system over at The Book blog recently and said that through testing they're confident that the tracking is within 2 inches with regards to a pitcher's release point and within 1" as the ball crosses the plate. Even with a 1" margin of error and remembering that the data points I'm using are much smaller than an actual baseball, that's still a fair number of pitches that Maddux seems to get the benefit of the doubt on.



Against righthanders he seems to catch more of the plate and interestingly doesn't seem to pitch as much down in the zone.

The Gospel of Pronation

At last year's SABR convention I had the opportunity to listen to Mike Marshall, former Cy Young winner and PhD, preach the gospel of injury avoidance through his rather unique pitching mechanics and instruction. At the time I noted how he's not taken seriously by the industry and now Jeff Passan over at Yahoo Sports has written an excellent article on Marshall (or "Doc" as his disciples call him) and his Pitching Research and Training Center in Florida north of Tampa. What's particularly interesting is the accompanying video of Marshall protege Jeff Sparks who once pitched for Tampa Bay and who still travels to Marshall's complex in hopes of catching on somewhere.

In the video you'll see the extreme pronation of the wrist in both the fastball and curveball motions as well the screwball which Marshall preaches will effectively eliminate elbow injuries requiring Tommy John surgery. The video is high speed and so shows the motion pretty clearly. You'll also want to check out Passan being interviewed by Will Carroll on BP Radio from last weekend.

In the end it would be interesting if a team would send a few borderline pitchers Marshall's way in order to see if there's anything to his claims about injury avoidance and additional velocity. Those who attend his center are certainly true believers but the rest of us won't be until there are some documented successes.

Quick Workers and Human Rain Delays Followup

My column this morning on Baseball Prospectus is a follow up to a post last week on fast and slow workers. Specifically, I was interested in the question of whether pitchers who work more quickly reduce the number of errors committed behind them as the common wisdom would indicate.

Although it's difficult to identify quantitatively which pitchers are sloths (the slow ones like Steve Trachsel) and which are humingbirds (faster workers like Bob Gibson) my attempt using anecdotal evidence couldn't find any statistically significant difference between two groups of 10 pitchers encompassing over 40,000 innings pitched since 1970. At first glance it was the sloths who seemed to suppress the number of errors. However, pitchers who worked faster in my sample were more likely to be ground ball pitchers and so that fact had to be corrected for since groundballs are more likely to produce errors than fly balls or line drives. I also used a subset of the two groups whose performance was almost equivalent to remove the bias that good pitchers introduce by being able to suppress errrors and unearned runs.