Where the wild things are

I’ve been thinking about minor-league numbers again and how to make sense of them. Earlier this year, Justin Inaz discussed run-scoring environments of different leagues. Not long before that, Harry Pavlidis examined ground balls and their impact in various leagues. And earlier still, David Gassko, in considering changes in league quality, referenced Bill James’ observation that among the factors that indicate quality is fielding percentage (see page 876 of The New Bill James Historical Baseball Abstract for a more complete list of factors).

The fielding issue is something I noticed a few years ago while researching the Ducksnorts 2008 Baseball Annual. In attempting to account for different levels of offense in different leagues (much as Justin did in his fine article), I found it odd that in 2007, teams in the Arizona League scored a lot more runs than their MLB counterparts despite not hitting all that well. So I dug a little and found this:

          R/G   BA  OBP  SLG FPct UER/9 ER/R
AZL 2007 5.66 .264 .350 .377 .952  1.23 .787
MLB 2007 4.80 .258 .336 .423 .984  0.36 .925

We would expect, based solely on their respective BA/OBP/SLG lines, MLB teams to slightly outscore AZL teams. However, in 2007, the opposite was true, by 0.86 runs per game. Why? The clues lie in those three rightmost columns: AZL fielders converted far fewer batted balls into outs, resulting in 0.87 more unearned runs per game than were allowed by MLB pitchers, thus accounting for most of the difference.

This phenomenon isn’t isolated to a single season. From 2007 to 2009, there was a predictable progression whereby the number of unearned runs allowed per game decreased with each advancing level:

Level  UER/9
F-Rook  1.30
Rook    1.06
ss-A    0.86
A       0.76
A+      0.71
AA      0.60
AAA     0.55
MLB     0.37

There are many possible reasons for this. I noted two of them (experience, playing conditions) in Ducksnorts 2008:

Players who have less experience (and who are honing their skills under less optimal conditions) tend to make more mistakes.

My goal in pointing this out was to provide context in evaluating Padres prospects and to explain why I gave stat lines for players at low levels despite the questionable reliability of said lines (to say nothing of said prospects):

I provide numbers on short-season players for the sake of completeness but request that you resist the urge to use them as proof of anything. Between the small sample and uneven level of play, to do otherwise is to beg for trouble.

That’s a lot of background information, but it’s important to introduce this stuff so that we remember a few key points going forward:

  1. The game of baseball as played in the low minors is a very different animal than the version pumped into millions of living rooms each week.
  2. People typically acquire and retain skills as they gain experience (and age, although there is a point at which the opposite becomes true).
  3. Defense and pitching are closely related.

That last point is a hunch and by far the weakest part of my argument. I won’t even get into it here other than to say I suspect it’s no accident that teams whose fielders have trouble converting batted balls into outs also have pitchers who have trouble finding home plate. Or maybe it’s the other way round, I don’t know. If a causal relationship exists, my guess is that poor control leads to poor defense.

Setting aside such speculation, the question I found myself asking was, “What is the difference in a pitcher’s control at various levels of play in professional baseball?” So I gathered data (from 2007 to 2009):

Level   UBB% HBP%  WP%   WE%
F-Rook 10.28 2.43 3.34 16.06
Rook    8.59 1.76 2.50 12.85
ss-A    8.87 1.63 1.95 12.44
A       8.11 1.37 1.80 11.28
A+      8.36 1.28 1.50 11.13
AA      8.78 1.13 1.23 11.14
AAA     8.34 1.02 1.19 10.55
MLB     8.03 0.89 0.83  9.75

Most of this should be straightforward: UBB% is unintentional walks divided by plate appearance times 100. HBP% is the same for hit by pitches, WP for wild pitches. WE% (Wild Events) is the sum of those three and is best said in a high-pitched voice while waving one’s hands in the air like one just doesn’t care.

As with unearned runs, we see a fairly consistent pattern from bottom to top. The only exception is that Double-A and High-A are reversed. On closer examination, however, we see that the two levels are very close and that the culprit is an elevated walk rate at Double-A. If we eliminate walks (in which sometimes pitchers miss by a little) from the equation and focus solely on the other two events (in which case they typically miss by a lot), we find again our straight progression:

Level  HBP+WP%
F-Rook    5.78
Rook      4.26
ss-A      3.58
A         3.17
A+        2.77
AA        2.35
AAA       2.22
MLB       1.72

Removing walks has another effect that I hadn’t considered until a friend brought it to my attention in conversation. The quality of umpiring in the minor leagues is haphazard as well and introduces another variable, i.e., that of a third-party’s judgment. The other two components are more pure in that regard. When a pitch gets away, it gets away.

At any rate, here are the numbers for all North American affiliated leagues from 2007 to 2009. I’ve included H+W% for the sake of completeness:

Lg  Level   UBB% HBP%  WP%   WE% HBP+WP%   
DSL F-Rook 10.61 2.41 3.48 16.49    5.89
VSL F-Rook  8.90 2.54 2.76 14.20    5.30
APP Rook    8.09 1.69 2.45 12.23    4.13
AZL Rook    9.17 1.87 3.09 14.13    4.96
GCL Rook    8.66 1.90 2.27 12.83    4.17
PIO Rook    8.49 1.54 2.36 12.39    3.90
NWL ss-A    9.46 1.84 2.29 13.59    4.13
NYP ss-A    8.51 1.50 1.74 11.75    3.24
MWL A       8.17 1.35 1.71 11.23    3.07
SAL A       8.07 1.38 1.87 11.32    3.25
CAL A+      8.36 1.32 1.64 11.32    2.96
CAR A+      8.64 1.30 1.56 11.50    2.86
FSL A+      8.17 1.22 1.32 10.71    2.55
EST AA      8.69 1.07 1.15 10.90    2.21
SOU AA      8.98 1.13 1.27 11.38    2.40
TEX AA      8.70 1.21 1.29 11.20    2.50
INT AAA     8.05 1.00 1.04 10.10    2.04
MEX AAA     8.44 1.18 1.35 10.97    2.53
PCL AAA     8.49 0.93 1.20 10.62    2.13
AL  MLB     8.02 0.89 0.85  9.77    1.74
NL  MLB     8.03 0.89 0.81  9.73    1.70

This table comes in handy when you want to say things like, “Batters get hit by pitches about twice as often in the GCL as in the PCL,” or “DSL pitchers throw more than four times as many wild pitches as their AL counterparts.” When might you want to say such things? If you have to ask, then clearly you are being invited to the wrong parties.

A Hardball Times Update
Goodbye for now.

On a more serious note, we see that even within levels, gradations exist: Arizona League pitchers are wilder than their Pioneer League counterparts, while California League pitchers are wilder than their Florida State League counterparts. To a lesser degree, this occurs between levels as well; note the difference, e.g., between the Northwest League and the Pioneer League. I don’t know why these differences exist, but they do.

Practical applications of all this? Well, I’m working on that. Actually, I’m not; I was hoping you might. Beyond the fact that it reminds us that context is important, I’m not sure there are any practical applications… yet.

That’s fine; often we acquire first the knowledge and then the understanding of how to use that knowledge. In this respect, we are like the young pitchers in the DSL, who acquire first the ability to throw baseballs and then the understanding of how to use that ability.

References & Resources

Influence and Inspiration

Data

Baseball-Reference:


4 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Geoff Young
13 years ago

Alternatively, I will have to cut myself off a bit sooner… thanks; it’s been fixed.

You make an excellent point about the quality of catching; I had not considered that.

kds
13 years ago

I think that all the stats you look at have a context based on the hitters to a varying degree.  Probably least for HBP and most for walks, with WP in between.  In one of his books, (NBJHBA?), didn’t James tell about a pitcher promoted from the low minors to the majors,(years ago), who hit a lot of batters in the minors but not in the majors.  About which he said that major league batters must have better reflexes than class C or whatever he came from.

LynchMob
13 years ago

re: catchers … I was thinking you could add “passed balls” as a “wild event” … might help the party conversation also …

Peter Jensen
13 years ago

Nice article Geoff.  At your next party you may want to change your bon mot to “Batters get hit by pitches half as often in the PCL…”.  Part of what you are seeing in decreased WPs may be do to better catching at higher levels as well as better pitching.  Also, the increased walks in AA may be due to the promotion of hitters with better strike zone recognition and pitchers increased use of descretional walks.