Every June, Major League Baseball holds its first-year-player draft, the principle way that amateur players enter the MLB system. Teams are allowed a certain number of picks, selecting players in an order determined by the previous year’s record; picks can be traded between teams to change this order. I became curious about how geography influences the draft. Do different teams focus on different regions, and does this change over time? Have the primary sources for players changed over time? Does this affect the quality of the results?
In 2014, The Hardball Times published an in-depth analysis of Geographic Biases in the MLB Draft, which focused on the regional valuation of players:
The variable we will key in on is the difference between the percent share of players drafted and the percent share of major leaguers. In a world where teams draft perfectly, this difference would be zero. That is, in each state players of a certain subgroup are drafted in the same proportion that they are represented in the big leagues. Any deviation from zero represents an inefficiency.
The authors concluded that:
…there is a real, persistent bias in the draft that undervalues players from baseball power states and overvalues players from non-baseball power states…a shift needs to take place, as the imbalance is too large to ignore.
The full article is worth reading, but its data presentation is mostly tabular. It doesn’t present the full geographic context of the draft, and doesn’t consider how individual teams’ draft picks compare to the league trend overall. In 2015, MLB.com published a simple infographic showing the total players drafted from each state, but this didn’t provide any useful analytical content with reference to time spans or teams. Thus, I decided to explore the geography of Royals drafting over time using both maps and graphs. Although the latter are more precise, the former can give a more intuitive feel for data patterns.
First, the broad context: here’s the annual average percentage of players drafted by state from 2017 back to 1998 (when the league expanded to 30 teams; see below for notes on methods). California, Texas, and Florida are clearly the powerhouse player factories, with a second tier of Arizona, North Carolina, and Georgia. Population is not the determining factor here; although the three most populous states are also the top three draft sources, the next four (New York, Pennsylvania, Illinois, and Ohio) do not stand out as draft sources while Georgia and North Carolina round out the top ten.
How have Royals draft picks compared to this MLB average? Simply making the same map as above, but for the Royals, would be misleading because the long-term average of a single team obscures the changes in management and draft philosophy contained within that period. For example, the Royals have had several general managers in this period, from Herk Robinson (1990-2000) to Allard Baird (2000-2006) to Dayton Moore (2006-2017). The Royals’ scouting directors have changed on a different timeline, from Deric Ladnier (2001-2008), J.J. Picollo (2009-2010), and Lonnie Goldberg (2011-2017).
To dig deeper, it’s tempting to focus on the Dayton Moore era, which in terms of drafting starts in 2007 as Moore was hired immediately following the 2006 draft. Certainly, Moore has changed the Royals’ competitiveness; going back to 1998, the Royals never finished higher than 3rd in their division until the 2014 (2nd) and 2015 (1st) teams made back-to-back World Series. But a look at the raw data suggests that it’s not Moore who has had the most direct effect on Royals’ drafting, but the people he’s chosen to hire. Below are the annual draft picks from the top two tiers of “power states” for both MLB and the Royals:
To my eyes, there’s a noticeable shift in geographic focus over time, but it really takes hold in 2009-2011, not 2007. That’s when Ladnier (Baird’s scouting director) was replaced by Moore’s own choices (Picollo and then Goldberg). So what does the geographic distribution of Royals’ draft picks look like from 2001-2008 (Ladnier) compared to 2009-2017 (Picollo/Goldberg)?
It appears that under Moore’s influence (in comparison to Ladnier’s previous approach), the Royals have de-emphasized four of the six power states in favor of more widely-distributed regional drafting. Under Moore, the Royals’ draft picks from all six power states are more tightly clustered at or below the MLB average (see graphs above), whereas Ladnier seemed have a wider year-to-year range. Moore seems to have placed a few scouts on a steamboat chuffing up the Ohio River and takes Missouri and most of its border states more seriously than Ladnier did. To check these conclusions, I graphed Missouri, Arkansas, and Kansas in the same way as the power states above, with clear results:
There are several possible explanations for all this: a desire to be different and look for diamonds in the rough, a refocusing of scouting resources closer to home, or mostly random chance in a relatively small sample size. The latter is certainly possible, as most of these changes represent just a few players per year, and various studies have suggested that drafting below the first few rounds is mostly a crapshoot. This story may be a tale told by an idiot, full of data and graphics, signifying nothing.
However, it’s interesting to me that the Hardball Times study specifically argued that teams are undervaluing players from power states, while the geographic evidence suggests that the Royals have made a point of doing that, especially under Dayton Moore. Is it coincidence that, when Royals Review looked at the success of Lonnie Goldberg’s drafts last June, the results were underwhelming?
Just comparing first-round picks, Ladnier was responsible for Zack Greinke (Florida), Billy Butler (Florida), Alex Gordon (Nebraska), Luke Hochevar (Texas - he pitched for an independent league team there after attending the University of Tennessee and high school in Colorado), Mike Moustakas (California), and Eric Hosmer (Florida), all of whom reached the World Series with the Royals except Greinke (who was traded for two starters on the 2014-2015 teams, Alcides Escobar and Lorenzo Cain). In other words, Ladnier’s style of drafting was responsible for six of the starting nine players in the 2014 World Series, just from the first round alone. Picollo/Goldberg’s picks (and thus Moore’s) haven’t come close to realizing the potential or success of Ladnier’s core. Obviously, geography isn’t the only factor here, but the parallels are intriguing and I’m curious what readers think.
If there’s interest, in a future story, I’ll use similar methods to explore how Royals drafting under Dayton Moore has compared to other teams known for their recent drafting prowess/success, such as the Cardinals, Astros, Cubs, and Dodgers.
Notes on methods
- All data came from the Baseball America draft database. I built my own spreadsheet of all annual draft results by pick#, position, player name, institution, and location, from which I can calculate a wide variety of comparisons.
- All maps were developed using QGIS, an open-source mapping platform that allows for robust geographic analysis far beyond the basic demands of this work but makes basic data mapping a breeze. Basic data analysis was conducted in Microsoft Excel.
- All data are presented as percentages rather than raw numbers because the total number of draft picks changes over time. For example, prior to 2012, the draft lasted 50 rounds with an average of ~1490 players taken per year (through 1998). From 2012 onward, the draft was shortened to 40 rounds with an average of 1219 players taken per year. This loss of ~271 players/year makes direct comparisons of raw numbers before and after 2012 difficult. Using percentages eliminates this problem because it effectively normalizes the data; just recall that a given percentage from 2012 onward represents a slightly higher number of players compared to 1998-2011. Roughly speaking, one player equals 2-3% of the annual total for a given team.
- Although the draft draws from Canada and occasionally internationally, these non-US players represent too small a fraction of the total to be worth mapping. Canadian players have averaged 1.4% of the annual draft from 1998-2017, never surpassing 2%, and the occasional German, Dutch, Australian, Cuban, or Dominican player(s) are a rounding error. However, they are included in all calculations, as are Alaska, Hawaii, and all US territories.
- All location data are by institution (school or independent team) rather than birthplace. Not only are these the data available, they also better reflect how and where players develop. Birthplace may have little to do with skills, but the high school or college are more representative of where young players develop.