In his efforts to objectively settle debates about athletic prowess, one blogger has developed statistical similarity networks for NBA players and Major League Baseball batters.  By finding each player’s ratio for all statistical categories (e.g. Free-throw Percentage/Turnovers per game), then normalizing the results in each category, the author found the Euclidian distance between each pair of players.  In resulting network, the nodes represent individual players and their career statistical set, while, and edges directed edges connect any player to the player whose competitive output is most similar (the least sum of squared differences between their statistical ratios). 

The info network graphs visually display components of similarly effective players.  Some of the results are quite intuitive, such as the proximity between steroid-linked power hitters like Barry Bonds, Jose Canseco and Gary Shefield.  However, this method also helps uncover pairings such as Joe Carter and Nomar Garciapara, whose different positions, size and race make them a non-obvious statistical duo that subjective analysis would probably overlook. 

            This methodology provides a compelling approach to verify claims of similarity and to discover new matches.  One could extend the system to comparing non-individual actors such as universities, sports teams or even national/regional economies.  If combined with time-series analyses, this type of info network could also illuminate the development of certain types of components or “authorities,” nodes with a concentration of inward connections, which represent the leaders or most-

emulated actors in a given field.