Skip to main content



The Best Tennis Players of All Time

In class, we have discussed Google PageRank, which was first invented in 1998 to analyze connections among web sites for the purpose of ordering search results. However, PageRank can be used for many other reasons as well. Whenever there exists a graph, which is math concept representing the relationship between a set of objects, PageRank can order these objects. One paper, written by Filippo Radicchi in 2011, utilizes the PageRank algorithm in order to determine the best professional tennis player of all time.

Using data from the Association of Tennis Professionals (ATP) of every tennis match from 1968-2010, Radicchi creates a network of contacts between all tennis players. Every time a player beats an opponent, a weighted edge is created between both players. These edges can become quite complex, since it is more impressive to win against a great player than against a less relevant opponent. According to Radicchi’s system, the ranking of players is not only related to the number of victories only, but also to the quality of these victories. Using “prestige scores,” which are analogous to PageRank scores, Rachicchi was able to rank the best players of all time.

The top 10 players by prestige score are as follows:

Screen Shot 2015-10-20 at 7.24.12 PM

Compared to two other traditional tennis ranking methods, Radicchi’s method is more accurate due the fact that it does not require arbitrary external criteria. However, the prestige scores still lead to a certain level of bias. Generally, the prestige scores favor players with the most number of victories, and therefore players who have already ended their career are ranked higher than active players who have not played all their career matches. However, this bias can be suppressed by only considering matches played in the same year.

The underlying idea in this type of analysis is that a fan may follow a player until another player beats them, at which point the fan will pick a new player. This is similar to how a person will surf around on the web.

Link:

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0017249

Comments

Leave a Reply

Blogging Calendar

October 2015
M T W T F S S
 1234
567891011
12131415161718
19202122232425
262728293031  

Archives