THE BOOK--Playing The Percentages In Baseball

If you read the full report you will find that in many ways the baseballs did not meet MLB specifications. However, the baseballs did not exceed the maximum allowable coefficient of restitution at the 58 MPH impact speed (although they were very near the maximum).

No baseballs from earlier seasons (pre-1993) were tested, so it isn’t surprising that they found little or no difference between the MLB baseballs. I strongly suspect that before 1993 the MLB baseballs were similar to the current minor leage baseballs, which fly about 98% of the distance. While that doesn’t sound like much of a change it is enough to turn a warning-track fly into a HR.

The 2000 report can be found here:

http://webusers.npl.uiuc.edu/~a-nathan/pob/baseball-study-report.doc

Great stuff, thanks!

The hitting machine measures the ball exit velocity after a precisely controlled collision occurs with a bat and baseball each traveling at 70 mph.

Interesting.  So they did a test of a Jamie Moyer changeup thrown to a Juan Pierre swing.  Why not test it at 20mph, so we can see how my kid would do?

The Major League balls are manufactured in Costa Rica and have a compressed cork sphere per the specifications.  The Minor League balls are manufactured in China and have a cork center as specified in “1996 Minor League Baseball Proposal”.  This cork center is the likely source for the decrease in performance, which results in a comparable Minor League ball hit of 391.8 ft under the same conditions as the Major League balls.  Small samples of 1998 MLB baseballs were also tested.  The 1998 MLB baseball had a comparable batted-ball distance of 400.5 ft. 

That looks like a substantial difference to me. 

And in fact Sherwood says:

Unlike the Major League pill, which is a compressed-cork center, the Minor League pill is constructed with a pure-cork center.  This pure-cork center results in a lower performance for the Minor League ball with batted-ball distances approximately 8 feet shorter. 

None of the testing was prior to 1993, which is the point when there’s been a substantial jump in runs per game, which has been sustained since.

Hittrackeronline.com notes HRs that only clear the fence by 10 feet or less (vertically or horizontally, which I think there’s a big difference between the two, but I digress).

If we take away the “just enough” HRs from the players, Thome goes from 35 to 20 last year.  ARod goes from 54 to 41.  Fielder, who hit the ball harder than anyone last year, goes from 50 to 38.  Brandon Phillips goes from 30 to 16!!!

Makes Wade Boggs and the other skinny dudes of the 1980’s seem a lot more impressive for cracking 20 or 30 HR.

Mike, that’s very interesting. 

Perhaps I can ask Greg, if you haven’t extracted the data.  Here’s my question: presuming that the HR per contacted ball rate from 1989-1992 is the baseline, how many feet would you have to reduce on flyballs in 2007 to turn the “just over the fence HR” into “warning track flyballs” so that the HR per contacted ball rate in 2007 matches the baseline?

If the answer is anywhere close to 10 feet, that would certainly be very interesting.

I don’t care whether baseballs fall into the “corrider” of accepted specifications that MLB requires.  I care about two things (actually only the first one):

One, for every year, I want to know how far/fast a ball is expected to fly (using some constant parameters of course), so that I/we can normalize hitting every year. (It would also be nice to know if anything else about the ball changes year to year, such as the height of the seams, the construction or feel of the leather surface, etc.)

Two, I want to know what happens if you test a current ball 5 or 10 (or 20) years later.  IOW, we don’t know that testing a pre-92 or 93 (don’t forget that HR and run scoring jumped radically in two consecutive years) is going to tell us anything unless we know how time affects a ball (e.g. where is the ball stored, what is the humidity content?).  IOW, if a 1990 ball tested now suggests that it is not as lively as current balls, is that because it was not as lively back then or is that because it has been sitting around for 17 years?

Even the current balls are in question, since you don’t have a “chain of custody”.  You have no idea where the balls have been, on their way to the scientists.  The best way for the scientists to test the balls is to physically be at home plate next to the umpire as they receive the balls from the ballboy, taking a couple away each inning, and then containing the ball in something that will preserve the ball.

Likely overkill, but then that leads into what mgl is saying.  If that’s what you really have to do, then the divergence from what they really did for this test (1999,2000 balls) is really magnified compared to what they would end up doing with donated 1990 balls.

Jay Jaffe has written a few things on the subject at BP, one of which I can find at the moment....

http://www.baseballprospectus.com/unfiltered/?p=128

One other study is here(with pictures):

http://tinyurl.com/yfnsnq

Among other things, it states that there is an extra piece of rubber around the core which is not included in the specs.

An 8 foot difference in fly balls of home run depth is massive!  In my THT Annual article, I described how I found that a 3 foot reduction in average fly ball distance from 2006 to 2007 knocked home runs down by 8%.  An 8 foot difference would be even bigger.

In 2006 there were 1454 homers in the “Just Enough” category, which means clearing the fence by approximately 10 feet or less.  They are spread fairly smoothly from 0-10 feet of clearance, so knocking 8 feet off every one would make 80% of them stay in the park.  80% of 1454 is 1163 fewer homers.  There were 5386 homers in 2006, and 1163 fewer makes 4223 homers, or a 21.6% reduction.  So, hypothetically, if MLB used the MiLB baseballs, we’d see 21.6% fewer homers…

Here’s some more data to chew on: the 2007 home run database, with a column added showing how far past the fence each one landed (note: the handful of negative numbers are either inside the park homers, or ones where the fence took an angle and my quick and dirty method of comparing the trajectory to the fence distances was off a little)

http://www.hittrackeronline.com/Master_Upload_file_2007_final_w_distancepastfence.xls

Here also is the complete data set for batted balls by Andruw Jones and Torii Hunter in 2007.

http://www.hittrackeronline.com/Hunter_Jones_2007_Raw_Data.xls

In 1990, there were 16.7 HR per AB-K

In 2007, it’s 21.9.

In order to matchup, you have to reduce HR by 23.7%.

Using Greg’s data above, 8 feet = 21.6% reduction, so a 24% reduction means 8.8 feet.

And what did the good doctor find?  400.5 feet minus 391.8 feet = .... 8.7 feet.

Wow.

That gun is smoking, I think…

But haven’t the parks gotten substantially smaller since 1990 as well?  Wouldn’t you need to do a similar analysis of the just enough home runs using the dimensions of the 1990 parks?

Some thoughts on HR rates through history:

Relatively minor changes to the baseball can have a considerable impact on batting stats. The change from a rubber core to a cork core led to a flurry of .400 hitters in 1911-1912, along with an increase in all extra-base hits. The outlawing of the spitball, cutball, etc. created the batting boom of the 1920s. A change in the cover of the NL baseball allowed the entire NL to bat over .300 in 1930, and a subsequent change made the NL much more pitcher friendly compared to the AL from 1931 to WW2. Ballparks which hosted both AL and NL teams usually saw considerably more HRs hit in the AL games, indicating the AL baseballs flew farther.

Qualities influencing ball flight can be grouped into two categories.

1. Resiliency
2. Aerodynamics

Category 1 covers things such as the properties of the materials used in making a ball and variables in the manufacturing process. Category 2 covers the size, shape, and roughness of the ball, which determines how much drag slows it down.

The 1911 and 1993 increases in offense are almost certainly due to a livelier core. The Reach baseball, publicly introduced in the 1910 World Series (and perhaps tested in league games before then), was the first to use a cork core. The 1993 change was unannounced (and possibly an unintentional byproduct of a change in either the manufacturing process or materials) but can be inferred from the leap in offense during 1993-1994. Researchers at the University of Rhode Island revealed a much livelier core in samples donated for testing (from the 1995 and 2000 seasons). It is entirely possible that other modifications were made over the years, of which we have no knowledge. CAT scans of baseballs from various seasons suggest several different cores may have been used. The most common assumption for peaks and valleys in offense has to do with the tension of the yarn wound around the core. this is another potential variable. Incidentally, this may also affect the aerodynamics. A tightly-wound ball would probably be a bit smaller, reducing drag. The height of the seams and stitching would also affect drag.

While researching past seasons I find it interesting that, in the absence of rule changes, BABIP rates are generally much more stable than HR rates. Even in the terrific HR season of 1987, BABIP increased only slightly. This suggests to me that much of the historic variance in HR rates is due to changes in the aerodynamics of the ball. Where we have confidence that the baseball itself was made livelier we find a large change in BABIP.

BABIP jumped between 1992 and 1994, as did the HR rate.  I posted a blog entry on it several months ago.

Tango,

Exactly. The big jump in BABIP without a rule change, as in 1993, is a very clear indicator of a change in resiliency. Batted balls increased in velocity, making them harder to field. If the HR rate increased without a corresponding jump in BABIP that would indicate flyballs were carrying deeper (or the fences were closer), but without additional velocity.

I’m sure the HR rate is an inherently “noisy” stat. But when looked at over a period of many years you find stretches of stability mixed with anomalous seasons, and other periods when the HR rate fluctuates quite a bit. If we have confirming evidence, such as a change in BABIP or the rate of other extra-base hits, then we pretty much know something was up with the baseball. 1993, although not a historic HR season, was when all the indicators shifted in the batter’s favor. That shift intensified in 1994, then established a new plateau.

If HR increase we may not see a change in BABIP because balls that used to be hits in play (ie doubles off the wall) are no longer in play.

What could you do to make a ball carry deeper without additional velocity?  The wind can affect that but I don’t think the ball can be changed in such a way.

Hot days, no?

You can also try comparing to minor league changes, since they both play at the same time.

I don’t think you can make the ball travel further on fly balls without increasing velocity, other than simply “more fly balls” which might be caused by a change in the strike zone, batter or pitcher approach.

Peter, #12, where did you get that (that parks have gotten smaller since 1990)?  That is trumpeted by the media all the time, but is it true?  Are you simply regurgitating what you have heard (even smart people and good and knowledgable analysts are not immune to that)?

If you read my article on park factors referenced in a recent thread, you will see that HR PF’s have gone DOWN at least since 1993 to the present.  We know they went up in 93 with the addition of COL and FLO (one a pitchers park and the other an extreme hitters park).  I am not sure about 90 to 93 without looking at my “park changes database.”

Now one reason for them going down from 93 to 07 is the humidor in COL, but everything counts.  My guess is that from 1990 to 2007, parks stayed around the same (in terms of HR factor), so I don’t think that has anything to do with the HR explosion and I don’t think you need to do much adjusting for parks when you are analyzing HR rate from 93 to the present.  In fact, if you look at my charts, HR rate in the NL, by virtue of the parks, has never changed by more than 2% in any year, and in all but one year by 1% or less.  In the AL, it went down by 4% in 2000 (up 1% in the NL) and up 2% in 2003 (and up .5% in the NL), but 1% or less in any other year.

So in 2000, we expected HR rate to go down by around 3% because of parks, and in 03, up by around 2.5%, in both leagues combined.  Other than those years, plus or minus around 1% or less.

Even in 93, with the addition of COL, you are talking about only 1% in the NL or .5% for noth leagues combined. In 95, with Coors Field as opposed to mile high, the HR rate went up again in the NL, because Coors was much more HR friendly than Mile High, especially in RF.

Checking my database, in 91, the WS moved their fences in, and Camden Yards opened in 92.  I think those are the only changes from 90 to 92.  Camden Yards is definitely more hitter friendly than the old Memorial Park.  So parks got smaller from 90 to 92 with two park additions/changes, but not by all that much.  As I said, with the decrease from 93 to 07 and the increase from 90 to 93, I think it is about a wash.

If the aerodynamic properties of the baseball change the ball may fly a different distance for an initial velocity and launch angle. In particular, if drag is reduced the ball will fly farther. The reason baseballs fly so far in Colorado is the thin air at 5000+ feet above sea level. Thin air=less drag. If you can reduce drag in any way baseballs will fly farther without requiring a change in resilience.

Drag can be reduced by:

1. Making the ball smaller
2. Lowering the height of the seams
3. Making the surface rougher to induce turbulence

Baseballs specifications have rather loose tolerances, so some variance in the size of the ball is acceptable.

MGL - There was a question mark at the end of the sentence in post #12 about parks getting smaller for a purpose; I have never looked into park sizes so I had no idea whether they had actually gotten smaller or not.  I still don’t.  I did read your paper on component park factors, but as you correctly noted a lot of other factors contribute to a park’s HR park factor other than the physical distance to the fences.  My comment was aimed to those previous posters who seeemed to be implying that they had found a “smoking gun” since the increase in HR’s due to compositional changes in the ball seemed to so closely match the percentage changes in the overall HR rate.  They may have found the predominate factor, but if there have been changes in actual physical distance to the fences that would also change the overall HR rate then the “smoking gun” would disappear, and we would be back to the real world where there are many factors each contributing to the overall rate sometimes in conflicting ways.

KJOK’s Parks Database can be downloaded from the Yahoo group KJOKbaseball.  It’ll give you all the year-to-year configurations of each park, ever.

Coming to this late, but since [#12] asked the question about smaller parks…

When I wrote the chapter for Will Carroll’s The Juice (published in March 2005, just days before Rafael Palmeiro wagged his finger in front of Congress) on this topic, I found that playing field dimensions had mostly gotten bigger between 1990 (the last year in which there were no new ballparks) and 2004, not smaller:

NL / 1990 / 2004 / change
LF / 331.3 / 333.0 / +1.8
LCF / 375.8 / 375.5 / –0.3
CF / 402.6 / 404.5 / +1.9
RCF / 375.8 / 379.6 / +3.8
RF / 331.0 / 332.5 / +1.5

AL / 1990.0 / 2004.0 / change
LF / 327.1 / 328.7 / +1.6
LCF / 378.1 / 377.9 / –0.1
CF / 406.1 / 403.3 / –2.9
RCF / 374.9 / 374.6 / –0.3
RF / 323.1 / 324.4 / +1.3

Rerunning the 1990-2004 comparison with KJOK’s park’s database (which ends at 2004), I get different values from that which I gathered via Ballparks.com, which may not have been the world’s most reliable source. Still, the net effect is towards larger parks. Combining the two leagues, we see the following changes:

LF:  +1.9
LCF: +2.7
CF:  0.0
RCF: +2.7
RF:  -0.2

I am very late to jump in on this topic, but I do have a few things to say.

1.  Regarding post 19, I would bet that lowering the height of the seams *increases* the drag.  The reason is the same as the third point made.  Namely, lowering the seams makes the ball smoother, which increases the drag.

2.  It is not easy to do experiments on old baseballs, since they are hard to find in untouched condition and hard to know how they have been stored.  Nevertheless, I have done some experiments on balls from the late 1970’s, measuring the COR at speeds up to 120 mph.  When compared to 2004 baseballs (tested at the same time by me), I saw no statistically significant difference between the two sets of balls.  However, I hasten to add that the sample size was small for both sets (6 balls each).  So, what does that prove, you ask?  Only my point that it is hard to test older balls--there just aren’t that many around!

3.  The MLB specifies the balls be tested at 58 mph, a speed that is way too low to be relevant.  What matters is the relative ball-bat speed, which can up upwards of 150 mph.  It is well known that balls that perform identically at 58 mph do not necessarily perform identically at higher speeds (this was very clear in the testing I did).  The technology certainly exists to test balls at higher speed.  Sherwood can do it in his lab at UML.  It only requires the will (by MLB) to do it.

4.  Moreover, the allowed range of COR at 58 mph is 0.514-0.578, a huge range.  If the 12% range were extrapolated to higher speed, that would result in a spread of about 35 ft in the distance of a long fly ball.  Just to be clear, all of these balls would be legal balls under current MLB test procedures.  It is interesting that in the 2000 tests at UML, all the balls were very tightly clustered together near the top of the allow range.  The same was true of the 2004 balls I tested.  They were very uniform.  So, it would appear that Rawlings knows how to make a uniform ball.  MLB could tighten the allowed range considerably and it would not be an undue burden in the manufacturing process.  Again, all it takes is the will to do so.

5.  Bob Adair in his book (The Physics of Baseball, 3rd ed, p. 97) estimates that an increase in fly ball distance by 1% would lead to a 7% increase in home run probability (presumably, HR/batted-ball-in-play).  So, for a nominal 385 ft home run, a 10 ft increase is 2.6%, leading to an 18% increase in the number of home runs.  These number are not so different from what one gets using Greg’s “just enough” analysis.  Adair also talks about other factors that could change home run probability--there are many, as others here have also noted. 

6.  Having said all of what I just said, I am not persuaded that today’s ball is “juiced” and that the juicing accounts for the increases in home runs.  I am not saying it isn’t true, only that I have yet to see scientific evidence that it is true.