FanPost

The Significance of Minor League K-Rates

Strikeouts in Prospecting
by RedSoxFaithful

I started thinking about doing a study on prospect strikeout rates earlier this week, when reading a comment thread between slamcactus and casejud about Nick Franklin that culminated in this comment. The argument was over whether or not strikeouts in the minors were a good thing for hitters, something that would go completely against conventional prospecting wisdom if it were true. There aren't a ton of studies out there (to my knowledge, at least) that explore minor league plate discipline as a predictor of major league success, and the more I looked into it, the more interested I became. I decided to perform a big-picture study on the subject, with an eye on certain developmental questions.


The Mysteries of Minor League Strikeout Rates

Of the areas I wanted to explore, the first seemed obvious. Are prospects with high minor league strikeout rates more likely to bust than their low-strikeout counterparts? Alternatively, is there a range of minor league strikeout rates in which prospects bust significantly more often? These questions should allow us to isolate the part that strikeout rates play in determining which prospects bust and which pan out.

There are some other interesting extensions that such a study allows us to explore, however. For example, among those prospects that don't bust, is a low strikeout rate in the minors a good predictor of MLB success? Do these non-bust prospects improve on their minor league strikeout rates once they arrive in the majors? If we can restrict the data population to include only successful prospects, it would be helpful to see if minor league statistics were at all predictive of how successful a given player was in the majors.

Answering these questions could help us understand exactly how strikeout rates and plate discipline in general fit into prospect development. By better understanding the significance of minor league plate discipline, we can more accurately evaluate prospective hitters, and make more informed decisions about their development.

Preparing the Prospect Data Pool

Originally, I had the idea of going through and looking at the careers of minor league players that crossed certain strikeout rate thresholds. The problems with this became obvious as soon as I realized that the vast majority of these players were not prospects, and it was hard to find minor league stats for entire leagues before 2006. Instead, I decided to use Baseball America's Top 100 database. I took every Top 100 list from 1990 to 2007, deleted the pitchers, and removed players that appeared on the list more than once. I also removed players without minor league data (this ended up including various Japanese players, John Olerud, and Dernell Stenson).

Ending up with a list of 524 hitters, I used The Baseball Cube (TBC) to add statistics for each player. There were a few limitations inherent in TBC, most notably the lack of Plate Appearances as a statistic. Unfortunately, however, TBC was the easiest site to navigate in terms of quickly finding minor league and major league statistics, side by side, for each player. It was still time-consuming, if only because I was manually copying statistics into an Excel spreadsheet for all 524 players. While the possibility for human error existed in the process, I did my best to double check the numbers for each addition.

Anyway, the statistics I included were as follows: Career MiLB AB, BB, K, and OPS; and career MLB AB, BB, K, and OPS. From there, I did my best to estimate PA, using the quick and dirty sum of AB + BB. I realize that this is nowhere close to a perfect substitution for PA, but for our purposes, it's not really important, anyway. For the remainder of this article, when I refer to PA, I'm referring to that sum. I used PA to calculate K% and BB% (just K/PA and BB/PA, respectively). This gave me a data pool of 524 prospects with 14 statistics to work with.

How Many Prospects Fail?

Career MLB At-Bats

# of Prospects

% of Prospects

0

36

6.87%

Fewer than 500

139

26.5%

Fewer than 1000

185

35.3%

Fewer than 1500

227

43.5%

Fewer than 2000

272

51.9%

The nature of this population allows us to make some general observations that might not necessarily relate to minor league strikeout rates, but are interesting for prospecting in general. The first and most obvious thing to check is how many of these top 100 prospects burn out. Just sorting the population by MLB at-bats, I was able to compute the values in the table to the right. I was surprised that only about 7% of prospects never played a minor league at-bat, though the 26.5% that played fewer than 500 makes more sense if you consider that most teams are highly invested in Top 100 prospects, and there's not much of a difference in terms of playing 500 at-bats or 0. The bottom line is that if your career ends with fewer than 500 at-bats, you probably didn't "make it" any more than someone who quit before they got 1.

Comparing Burnouts and Successes

Before I looked at the differences between our burnout and success populations, I needed to define the terms "burnout" and "success". This was tricky, but, using an extension of the "500 at-bats" logic in the preceding paragraph, I went with a simple restriction of 1500 MLB AB. That gives a given prospect about 3 full seasons to try and stick around with a major league club. So, for the purposes of this essay, any prospect with at least 1500 MLB AB is considered a success. Any fewer than 1500, and the player is considered a burnout. Obviously, this is imperfect, and a few (read: maybe half a dozen) players that cannot reasonably be called busts missed the threshold, but it works more than well enough for this paper's purposes.

Averages

Successes

Burnouts

% Difference

MLB AB

4031

452

-88.8%

MiLB K/BB

1.79

2.14

19.6%

MiLB BB%

10.0%

9.7%

-3.0%

MiLB K%

16.4%

19.5%

18.5%

MiLB OPS

.820

.791

-3.5%

This restriction provides us with two distinct populations of successes and burnouts. The first thing I decided to do with these populations was to look at their averages in significant statistical categories. Those results are in the table to the right. All of the findings are obvious, and support conventional prospecting wisdom. Obviously, players who walk more, strike out less, and have higher OPS in the minors are more likely to stick in the major leagues than their counterparts. That much was already known.

The interesting part about this table, though, is the amount by which the populations differ in specific categories. Surprisingly, it appears that successes and burnouts really don't differ that much in minor league OPS or walk rates. The most important statistics here seem to be the K/BB ratios and the strikeout rates. This finding supports the reasoning that led me to perform this study. Simply put: strikeouts seemed to be pretty important in prospect evaluation. Exactly how important they are remained up for debate.

So, Is Striking Out a Death Sentence?

Despite the sharp increase in strikeout rates between successes and burnouts, it didn't appear that burnouts struck out that much. 19.5% isn't a startlingly high rate- the MLB average via our formula was 18.9% in 2010. I decided to break down successes and burnouts by specific strikeout rate thresholds to see if prospects are various rates were significantly more likely to succeed.

What I found, shown in the table below, wasn't really that surprising. At every above average strikeout rate threshold, players were more likely to bust than be successful. There were some interesting quirks in the distribution here. Although more than half (51.9%) of successful players had an above average strikeout rate, that percentage drops sharply as strikeout rate increases. For burnouts, however, we only see sharp declines in percentage around 22%, which also seems to be the rate where success rates plateau in the mid-20s. Without treating the area between 20%-22% too scientifically, it appears that that would be the region where one should start seriously worrying about prospects' contact abilities.

MiLB K%

Successes

Burnouts

Success Rate

>16.4%

154 (51.9%)

163 (71.8%)

48.6%

>18%

109 (36.7%)

142 (62.6%)

43.4%

>20%

52 (17.5%)

109 (48.0%)

32.3%

>22%

21 (7.1%)

69 (30.4%)

23.3%

>24%

10 (3.4%)

34 (15.0%)

22.7%

Someone arguing the meaninglessness of minor league strikeout rates might suggest that players with strikeout rates worth worrying about will weed themselves out by failing at more advanced levels, anyway. This actually sort of echoed a concern I had with my data set. As I was adding players, I noticed that a lot of the busts were players who were in the minor leagues well into their 30s. I was worried about how this would affect the data set, because it wouldn't account for players who "learned" to limit strikeouts over time. I decided to test this by doing a simple correlation test between minor league strikeout rates and minor league OPS, and found no significant correlation between the two (-.004). When it comes to minor league OPS, it doesn't appear to matter whether or not a player strikes out too much. This is an interesting conclusion, because it suggests a sharper increase in difficulty from the minors to the majors than one might expect. To double check that increase, I also did a correlation test between minor league strikeout rates and major league OPS. This time, there was a fairly strong negative correlation (-.184). Players who struck out more in the minors tended to have a lower OPS in the majors, despite strikeout rates not affecting minor league OPS in any discernible way.

Strikeouts among Successful Prospects

The problem with that correlation test is that it includes a lot of players with small-sample-size OPS. While the conclusion still holds true in general (because our burnout population tends to have lower MLB OPS anyway), it might be more meaningful to look at how minor league strikeout rates have affected the MLB hitting ability of the non-burnout population. I ran a simple correlation test between MiLB K% and MLB OPS among that success population. This was where things got a little wacky, as there was actually a fairly strong positive correlation (.143). What this means is that successful players with higher strikeout rates actually had better OPS!

There are a couple of possible explanations for this. Anecdotally, I found more than a few players in the sample that were able to stick around for a long time (say, 6000 AB) with lower OPS (say, .695). These players tended to be defensive specialists that, for one reason or another, were never demoted due to a lack of hitting ability. What hitting ability they did have, however, tended to be contact-heavy. These players did not strikeout often, which may have skewed the correlation a bit.

Alternatively, there are a good amount of sluggers who were successful despite large amounts of strikeouts. Again, anecdotally, I've heard people mention the Ryan Howards and David Ortizes of the world when defending high strikeout rates for sluggers. Those people probably aren't wrong, but Howard and Ortiz also aren't good comparisons for many prospects. Those types of players are the extreme exceptions to the rule, and there are almost no prospects that can be reasonably expected to match their power ceilings.

A third potential explanation is that better hitters tend to take more pitches, which tends to increase strikeout rates. I'm not entirely positive this would show up in minor league strikeout rates, since theoretically, these are still-developing prospects, but it would lend a hand toward understanding the correlation. If we allow these three theories to compound, it goes a long way towards explaining why minor league strikeout rates are associated with better major league production.

Ultimately, I decided to look at this a different way to try and better understand this phenomenon. I sorted the successful players by highest minor league strikeout rates and took a look at their MLB OPS, and then sorted by highest MLB OPS, looking at their minor league strikeout rates. The two resulting tables are below. I cut off the highest MiLB K-rates at 22% (the top 21 rates), along with the top 21 MLB OPS, for symmetry's sake.

By Highest MiLB K%

By Highest MLB OPS

Name

MiLB K%

MLB OPS

Name

MiLB K%

MLB OPS

Russ Branyan

32.8%

0.822

Albert Pujols

8.8%

1.046

Wily Mo Pena

30.0%

0.754

Manny Ramirez

19.8%

0.996

Glenallen Hill

28.6%

0.803

Todd Helton

13.1%

0.978

Ryan Howard

27.8%

0.943

Frank Thomas

15.0%

0.974

Tim Salmon

26.8%

0.883

Joey Votto

22.4%

0.966

Tony Clark

26.7%

0.824

Larry Walker

23.6%

0.965

Preston Wilson

26.6%

0.797

Jim Thome

18.4%

0.962

Jack Cust

26.0%

0.823

Alex Rodriguez

18.0%

0.959

Dean Palmer

25.5%

0.796

Lance Berkman

17.1%

0.956

Ryan Ludwick

24.0%

0.804

Jeff Bagwell

10.4%

0.948

Derrek Lee

23.7%

0.863

Ryan Howard

27.8%

0.943

Larry Walker

23.6%

0.965

Vladimir Guerrero

9.9%

0.943

Felipe Lopez

23.4%

0.732

Miguel Cabrera

16.8%

0.942

Mike Cameron

22.6%

0.786

Chipper Jones

12.4%

0.941

Brandon Inge

22.5%

0.700

Carlos Delgado

19.3%

0.929

Joey Votto

22.4%

0.966

Mike Piazza

18.4%

0.922

Eric Anthony

22.4%

0.702

Ryan Braun

18.0%

0.922

Wil Cordero

22.2%

0.758

Prince Fielder

17.1%

0.921

Chris Young

22.2%

0.759

David Ortiz

21.8%

0.919

Geoff Jenkins

22.1%

0.834

Josh Hamilton

18.7%

0.916

Ricky Ledee

22.0%

0.737

Mark Teixeira

16.3%

0.912

Only three names appear on both lists: Joey Votto, Ryan Howard, and Larry Walker, and they all have had exceptionally rare skillsets. The very best hitters seem to excel at limiting strikeouts, at least to some degree. That being said, there are a lot of solid hitters among our leaders in minor league strikeout percentage. Of course, this is all just a long, less scientific way of saying the same thing that the correlation statistics told us: Among successful major league hitters, those that had higher minor league strikeout numbers tend to be better. Does that mean it's better to have higher strikeout numbers in general? No, of course not- it just helps us better understand what types of hitters (sluggers) tend to produce more.

The final thing I wanted to find out about prospect development with regard to strikeout rates was whether or not our "success" group improved their strikeout rate upon graduating to the majors. Thankfully, this was an easy question to answer, with three easy ways to go about it. First, I looked at the correlation between MiLB K% and MLB K%. As assumed, there was a very strong positive relationship between them (.776). All this meant was that as minor league strikeout rates rose, major league rates rose pretty closely with them. Next, I took the average change in K% from the population, which turned out to be slightly under +1% (.993%). This more specifically answered my question; it showed that players generally see an increase in their minor league strikeout rates upon graduation to the majors. That increase didn't seem to be particularly significant, however. To double check it, I ran an OLS regression on MLB K%, using MiLB K% as the only independent variable. The regression was significant (R2 = .501), and the model gave me the following equation:

MLB K% = 1.14961(MiLB K%) - 0.000801827

Again, this supported the findings that strikeout rates rose upon prospect graduation, though once again by a surprisingly small amount.

Limitations Inherent in This Study

There a couple of problems I had with this study that I wanted to mention. I don't think they significantly affect the relevance of the research, or the credibility of the conclusions, but they are worth mentioning.

First, as I mentioned earlier, the samples are a little inconsistent. In retrospect, I probably shouldn't have included the 2006 and 2007 prospect classes. There are at least a few players I counted that made their way into our "bust" population that have no business being there. Similarly, our imperfect and general definition of "success" means that more than a couple of players snuck into that group that shouldn't have. All in all, I think these effects balanced each other out. The success and bust populations are generally correct in being labeled as such.

Second, the nature of prospect development means that cherry picking strikeout rates in order to solely explain a player's success or failure can be somewhat unfair. That being said, though, the prospect population is normalized in the sense that they were all at one point rated in the Top 100, and therefore had some promising skillset. An extension of this problem is that some of those promising skillsets were heavily oriented towards defense, and certain prospects weren't expected to hit well regardless of production. Obviously, though, major league teams can't tolerate exceptionally low offensive production. Players would have to be so gifted defensively to force teams to overlook their offensive shortcomings that the sample size is likely insignificant.

While there are a couple of issues with the prospect pool to be considered, they are mostly negated or accounted for without the need for much concern. These limitations should mostly be irrelevant to the research's significance.

Conclusions:

Admittedly, there are not a ton of revolutionary conclusions here. Traditional prospecting wisdom is supported strongly, as it's abundantly clear that, ceteris paribus, high strikeout rates are not helpful in a prospect's development. It appears as though the success rates for prospect development drop sharply when strikeout rates hit about 22%. Furthermore, minor league strikeout rates are strongly negatively related with both major league at-bats and OPS.

However, among successfully developed prospects, minor league strikeout rates are positively related with major league OPS, indicating that the prospects that become the best MLB hitters tend to be sluggers with moderately high strikeout rates. A second point to note is the abundance of defensive-minded players who don't strikeout much, but also don't produce much offensively. Lastly, it's clear that players tend to increase their strikeout rates in the minors upon graduation to the majors by about 1%.

Again, the conclusions made here are by no means revolutionary. That being said, it's good to have the numbers to back up conventional wisdom, and I'm sure there are various extensions to this study that could increase its usefulness. For now, though, we can use the data to help make predictions about prospects going forward, and become a little more wary of high strikeout rates.

Trending Discussions