Anjela Govan, North Carolina State University

Ranking National Footbal League teams using pagerank and other matrix recipes.

Abstract:  The National Collegiate Athletic Association football does not a have a playoff system. Instead the college teams are ranked throughout the season based on their performance. At the end of the year the top two ranked teams play each other to determine which one of them can claim the title of National Champion. The ranking formula used is extremely complicated and accounts for numerous parameters which rate the quality of the teams. In fact, determination of these parameters is a science in itself. What makes a certain team better then the other? Is it coaching staff? History of the teams’ performance? Individual talent of the players? Some factors we might perceive as extremely important might have an insignificant effect on the performance of the team. How does one rank anything? One can use intuition, but the best and most consistent approach is to use mathematics. The branch of mathematics, matrix analysis, is the top choice for constructing the ranking recipes. One of the currently most successful ranking algorithms, called PageRank, has been developed by the internet search engine Google. PageRank is used to rank the internet web pages and it is characterized by it’s global approach to the solution. Google’s idea of web page ranking can be used in other applications. In this work we expand Google’s idea of web page ranking to ranking National Football League teams. We think of the teams as web pages and use the statistics of one football season to create the connections (links) between the teams. The objective is to be able to rank NFL teams by their relative strength. To evaluate the performance of the PageRank on the NFL teams we implemented a number of other ranking algorithms.

Advisor: Carl Meyer (NCSU)