Michael Laterza, Robert E. Lee High School
Abstract:
Expert analysts have studied the statistics of football in the National Football League for decades, often predicting players' statistics or a team's chance of winning. However, few studies have been conducted to find the individual statistics that lead to a team's points scored, a correlation which could prove valuable to the NFL community. For the experiment, data was collected from a reliable football archive website for various statistics, including points, yards, Passer Rating, field location, etc. Descriptive stats were calculated for the data, and the various statistics were tested for correlation with points scored. As predicted, the best correlation coefficients (R-values) were between points and Passer Rating, which is a scoring system used to grade how well a quarterback performs (0.67), and points and yards (0.71). Turnover differential had the highest coefficient (0.72), proving that it was the most important statistic in determining the outcome of a football game. Most statistics didn't correlate well at all, but those three stats were proven to have importance in NFL games.
Introduction:
Since its humble beginnings, American Football has been an American pastime. It is often considered the country's most popular sport, which explains why over 100 million people watch the Superbowl annually, the National Football League's (NFL) most prestigious event. Experts frequently analyze the game and the statistics surrounding it, and innumerable Americans rely on this data, whether they are keeping up with their favorite team or participating in a fantasy football league. American Football has become immensely important in the US, and in turn the populace has begun to look deeper into the game. Experts have not been entirely successful in using this data for predictions, however; many NFL game predictions are wrong, some by huge margins. This is because football has many confounding variables, which can easily alter the course of the game. A team scores points by moving down the 100 yard field to score points, but these “yards” and “points” can be earned in many ways. Teams can either score a field goal, 3 points scored via a kick through the goalposts, or a touchdown, 6 points and possibly more scored by moving the ball into the opponent's end zone, not to mention the rare safety, 2 points scored by forcing the opponent into their own end zone. These many scoring options create some of the inaccuracies in football game prediction. The 22 players on the field each have their distinct way to alter the score in their favor. The quarterback can either throw to his receivers for potentially huge gains, or hand the ball off to his running back for a more modest gain. Defensive players can force a turnover, which gives their team the ball, or a sack, which forces the offense backward into their own territory. If simply one player has a good day, or, on the contrary, gets injured, the entire game changes. This makes it very difficult to properly analyze the primary causes for team success.
In a study conducted by Brian D. Lyons (2011), a football player's success in the NFL was compared to his success in college and in the NFL Scouting Combine, which tests players' skill before the NFL season begins. He found that college performance had a greater effect on NFL performance, despite the Scouting Combine being closer to the start of the NFL season. His study succeeding in finding a notable cause-and-effect relationship within football, similar to this experiment.
This research answered the question of what statistics influence the game of football, or if any do at all. The null hypothesis stated that no statistics would be significant enough to be considered influential to the points scored in the game. The alternate hypothesis stated that some of the data would