This project aims to analyze over 10 years of NFL game data to determine the predictive strength data has on NFL games. Often times superlatives are used to describe certain finding in data. Points are made using absolutes, however, the game of football is not so simple. Therefore, this project was conducted to see if there were any way to reliably predict game outcomes purely based on chosen strategy.
This project is looking at the current views surrounding data analysis regarding American Football and attempting to view them in a more accurate light. In recent years, NFL teams have started taking hard stances and decisive actions based on larger data beliefs. However, some teams tend to stick too hard to these beliefs (such as running having little value, going for it on 4th down, etc.) and have not had the success that some would lead people to believe. This overreliance on flawed interpretations has led to multiple teams losing games and, in turn, missing the playoffs. This can have damaging effects on an organization and its staff. This report hopes to assist any coaches who have been overwhelmed by the recent wave of data analytics in football by recontextualizing both what is the most important areas for statistical analysis derived decisions and when to deploy them.
If data isn’t properly understood in its analysis, the team will suffer both on the field and financially. Having a background playing Madden at the highest level gives me a unique understanding of all the different data points being used in these calculations, allowing me to better analyze the data than most. In this project, I’ve reviewed over 400,000 plays from 2,500 games over 10 years to attempt to verify or deny some of these current beliefs and ultimately lead to an even more winning strategy.
