Skip to main content



Oracle of Baseball

http://www.baseball-reference.com/oracle/

I have linked to the Oracle of Baseball application from Baseball-Reference.com (linked above). The application forms the shortest chain between any two players in Major League Baseball history by virtue of common teammates. Thus, you select any two players, like Babe Ruth and Derek Jeter for example, and the Oracle of Baseball lists the shortest chain of players that connect Babe Ruth and Derek Jeter. In my example it takes 4 players (5 links) to connect Babe Ruth and Derek Jeter. That doesn’t seem like very many when you consider Babe Ruth’s last season was 1935 and Derek Jeter’s first season was 1995 (60 years apart). Another feature of the Oracle of Baseball is that it can give you the “connectedness” of any single player. The “connectedness” of a player is the average distance (number of teammates) between that player and any other player. The most connected player according to the Oracle of Baseball is Minnie Minoso with an average distance of 3.0127 teammates to any particular player. This application takes the basic graph theory we have learned in class and gives it a fun application. The Oracle of Baseball infrastructure was created by Patrick Reynolds who also maintains the similar and well-known Oracle of Bacon, which takes the same concept and applies it to actors and their co-stars.

The Oracle of Baseball is a direct application of the theory of Six Degrees of Separation of human connectedness, which posits that any two humans in the entire world are connected through five series of acquaintances. With the extensive record keeping of Major League Baseball, it was possible to create a fairly good estimated (if not perfect) universe of Major League players, past and present, that was necessary to run this sort of case study. In the network that the Oracle of Baseball uses, each player is a node and each teammate pairing is an edge. According to http://www.baseball-almanac.com/players/ballplayer.shtml there have been 17,744 unique Major League Baseball players from 1876 through the beginning of the 2013 season. This universe is obviously much smaller than the human world, but it is still interesting to examine.

There a few nuances of the Major League players universe that create some quirks in the experiment. First, up until 1975 Major League Baseball did not have free agency. So, before this time player movement between teams could only occur if an agreement was reached to trade players. In the years since the advent of free agency players can choose which teams to play for upon the completion of their current contracts thus leading to great roster turnover from year to year. Having this knowledge of baseball’s free agency history gives context to the network as it is currently constructed, possibly explaining why older players might have fewer average links than more recent players. Specifically, this nuance relates to one of the two factors that determine a player’s connectedness through the number of teammates he has had. One factor is career length; longer careers give more opportunity for various teammates. The other factor (which relates to free agency) is the number of franchises a player has been a part; the more franchises a player has played for the higher greater number of possible teammates he may have had.

There are a couple of ways to further define the network in this experiment based on what we have discussed in class. Something which the Oracle of Baseball does not account for is link strength. All links in the experiment are assumed to have equal strength. In reality, some players spend more time together as teammates than others and thus we could give weighted values to the links based on this notion. Another addition we could make to this experiment is to create links between players who were opponents of each other. We could do this by assigning negative and positive values to the links. If two players have been teammates their link will be positive. If two players were never teammates, but did play against each other, their links will be negative. This idea of “opponent links” relates to another nuance of the baseball universe. Up until 1997 teams in baseball’s two leagues did not play each other during the regular season, so the number of possible opponents has increased since 1997, possibly resulting in recent players having more opponent links than older players would have. All of these tweaks add information to the network making it more complex and interesting than its current configuration.

Comments

Leave a Reply

Blogging Calendar

September 2014
M T W T F S S
1234567
891011121314
15161718192021
22232425262728
2930  

Archives