Throwin' it Out There…

From Alison Brie to Zach Randolph, and everything in between.

MLB Roster Construction: Source Code

“You aren’t handsome enough to scout for us. You’re fired.”

2011’s much-hyped Moneyball, for all its Sorkin-ized flaws, gave the general public a shallow insight into the Oakland Athletics front office (for a scathing explanation, see former insider Keith Law’s review). Unlike the film’s basic depiction of scouting and sabermetrics as purely black and white-haired, in reality organizations’ scouting departments collaborate with new analytics to evaluate players. Many teams still lean heavily in one direction or the other. I was thinking about this and the allocation of scouting resources by traditional teams, which begged the question: theoretically, if a team relies mostly on old school physical scouting, do they have the resources to cover every player on every team, and thus are limiting their potential player pool? Conversely, do stat-focused teams have a wider pool of players to choose from because they target players mostly based on metrics (to quote Aaron Sorkin’s Billy Beane, “I don’t watch the games”) ?

If this rings true, not only should saber-friendly teams have a variety of players from across all thirty teams, but “old school” teams should have players they drafted or from within their own division and league- those they have a familiarity with from playing frequently. On a side note, I hate the term “old school” regarding baseball scouts – it sounds pejorative, but so does “traditional”, “stat-averse” or “mathematically challenged”. I personally lean on the saber side, because grit, guts and David Eckstein don’t help me win my fantasy league.

To start, the highly analytical teams needed to be weeded out from the non. Luckily, the sabermetrics capital of the internet, Fangraphs, featured a post last year breaking down the league into highly analytical, in between, and Old School categories. I took six of the most obvious stat-friendly teams (Athletics, Mets, Blue Jays, Mariners, Rays, Cubs) and six of the most old-fashioned (Twins, Giants, Rockies, Tigers, Dodgers, Royals). All these organizations jump out as extremes in their respective groups- Billy Beane had a movie based on him and most baseball fans are well aware of Royals’ GM Dayton Moore’s fondness for walk-allergic Yuniesky Betancourt.

Yeah, he swung at this.

Each team’s 25 man roster (including disabled list, as of  4/14/13) was broken down by source of acquisition – each player was assigned a category based on proximity of their previous team. A player could come from a division rival, a team in the same league, an interleague rival, a team in the other league, from the draft, or internationally. Here are the results:

(click to expand)

(click to expand)

(click to expand)(click to expand)

A notable difference between the two groups dictates that scouting-centric teams roster more players they’ve drafted (34% to 22%) than saber-friendly teams. Teams who depend more heavily on scouting do so for a reason- because they’re good at it. I’m not going to pretend to know every input into amateur draft/international player evaluation, but I would imagine scouts have a huge hand in those areas, whereas sabermetric teams analyze players at the professional level. The numbers in this admittedly small sample size do back this up.

Another significant takeaway from these charts is that statistically-inclined teams acquire significantly more (42% to 23%) players from the opposite league. Take a look at the Oakland A’s and the World Series Champion San Francisco Giants. The A’s acquired a whopping 63% of their roster from the National League, while the Giants only acquired 18% of their roster from the American League, excluding Bay Area neighbor the Athletics (see below).


(click to expand)

One of the positions where it’s perceived that teams have drastically changed their search criteria for in recent memory is relief pitching. Quality relief pitchers- whether it’s because of injury risks, prioritization of starting pitching or advanced matchup analysis by managers- seem to be ubiquitous and interchangeable these days (don’t tell the 2012 Mets that). The trend towards “smart” teams saving resources by going after extreme veterans (old bastards, to put it bluntly) and rookies, while old school teams pay for past performance in relievers seems to be in full effect:

(click to expand)

(click to expand)

(click to expand)

(click to expand)

The sabermetric teams’ chart looks nearly identical to the previous one based on entire rosters. On the other hand, the old school teams’ acquire 10% more relievers from other major league teams, at the expense of 6% of their own draftees. These teams target established veterans like Rafael Betancourt (COL), Jeremy Affeldt (SF), Joaquin Benoit (DET) and Brandon League (LAD). Meanwhile, sabermetric bullpens consist of rookies and much older veterans like Latroy Hawkins (NYM), Kyle Farnsworth (TB) and Darren Oliver (Brooklyn Dodgers). Their resurgence is fitting, with Jurassic Park’s re-release in IMAX…wait a second, those are seven names, SMALL SAMPLE SIZE! Lucky for this article, the numbers tell the same story.


Even though the average years of MLB service for relievers in both groups are virtually the same, for saber teams, the median is lower, but the range is higher and standard deviation a full year greater …zzzzzzzzzz. In English: this group has a lot more young (aka cheap) players, but also more extreme veterans- interpreted from the significantly higher range and standard deviation despite essentially the same average as old school teams. On the other hand, old school teams have less rookies, but the veterans are closer to their primes than the dinosaurs above- Hawkins, Oliver, and Jamey Wright (Rays) each have over 17 years of MLB experience. Simply put, the iPod selection in the bullpen for sabermetric teams is way more diverse than for old school ones.

One area saber teams have definitely tried to save money with veteran minimums and pre-arbitration rookies is indeed relief pitching. For all the fuss about roster construction since the Moneyball era started, an interesting note is the dichotomy of actual performance by these teams last year- the old school teams consist of both World Series’ teams and two playoff teams, while the saber side is only represented by two reigning playoff teams. Does it really make a difference whether teams use spreadsheets or radar guns? Stay tuned for the next post looking for which of these teams are actually spending their dollars most efficiently.

Feel free to leave any comments or questions below or @kramerkram. Any complaints please keep to yourself.


2 comments on “MLB Roster Construction: Source Code

  1. Enolazna
    April 18, 2013

    Awesome article. Looking forward to money inefficiency.

  2. Pingback: MLB Cost Per Win: Which Front Offices Save the Most Money? | Throwin' it Out There...

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s


This entry was posted on April 18, 2013 by in MLB and tagged , , , .
%d bloggers like this: