• Bruin Sports Analytics

The NFL Combine: Does It Predict Quarterback Success in the NFL?

By: Shayan Ghafoori and Andrew Schweitzer

Sources: Bleacher Report, Sporting News


Introduction


The controversy surrounding the validity of the National Football League (NFL) combine and its role in predicting a prospective player’s success is a debate so old that it has become a cliche. Yet, each spring, the NFL continues to conduct a series of standardized tests designed to assess the skills of young, aspiring professional football players. Despite already showcasing their talents throughout their college football careers, the combine allows players to present their athletic prowess in a controlled environment, potentially hurting or helping their draft stock. Given this, the NFL clearly views the combine as an--at least somewhat--influential metric in assessing a player's performance, but what does the data say? This research article primarily aims to assess the degree to which the combine predicts NFL performance. To remain focused, we specifically analyzed the combine statistics at the quarterback position, concentrating on eventual NFL quarterbacks who participated in the combine and were drafted from 2000-2020; our research question being: Is there a correlation between a quarterback's performance at the NFL combine and his success as a professional football player?


We chose to analyze possible correlations between NFL combine results and measures of NFL quarterback success. The combine results we decided to use in this study are the 40-yard dash, vertical jump, broad jump, three-cone drill, and 20-yard shuttle. Our criteria for measuring quarterback success are: games played, touchdown/interception ratio, and completion percentage.


Methods


We approached our question by forming a null hypothesis: Combine results are not predictors of success for quarterbacks in the NFL. Null hypotheses assume there is no relationship between two variables, in particular, controlling one variable will not affect the other. To prove whether we reject our null hypotheses or fail to reject, we calculated the p-value and compared it to a significance level of 0.05. The p-value tells us the probability that an observed difference may have occurred only by random chance. If our p-value is less than the significance level (0.05) then we can say that the correlation is statistically significant and is different from 0. This suggests that a relationship may exist since the correlation coefficient is at least different from 0. However, if the p-value is greater than the significance level then our correlation coefficient is not statistically significant and not different from 0. This may suggest that a correlation does not exist since it is not different from 0. The correlation coefficient is a value representing the strength to which the movement of two different variables is linked. A positive coefficient signifies that the variables move in the same direction; on the other hand, a negative coefficient signifies that the variables move opposite from each other.


Results


The first measurement of success we compared to NFL combine results was games played. We looked at the correlation coefficients for each comparison between games played and combine results; the strongest correlation coefficient found was games played and 20 yard shuttle (-0.07398524). To check if the correlation coefficient is statistically significant, we calculated the p-value and compared it to a significance level (0.05). We found a p-value of 0.3938 which is greater than the significance level, so we can conclude that the correlation coefficient is not statistically significant and is not different from 0.

The graph above is a linear model between games played and 20-yard shuttle time and as expected, the correlation between the two variables appears to be absent.


The second measurement of success we compared to NFL combine results was completion percentage. We looked at the correlation coefficients for each comparison between completion percentage and combine results; the strongest correlation coefficient found was completion percentage and weight (-0.2233517). However, we filtered out quarterbacks to only those who have played 16 or more games so that our results were not skewed. We found a p-value of 0.01462 which is less than the significance level, so we can conclude that the correlation coefficient is statistically significant and is different from 0. Although, by looking at the graph, it is difficult to identify a correlation.

The last measurement of success we compared to NFL combine results was touchdown/interception ratio. We looked at the correlation coefficients for each comparison between touchdown/interception ratio and combine results; the strongest correlation coefficient found was touchdown/interception ratio and 3-cone drill time (-0.1243057). Again, we filtered out quarterbacks to only those who have played 16 or more games so that our results were not skewed. We found a p-value of 0.2658 which is greater than the significance level, so we can conclude that the correlation coefficient is not statistically significant and is not different from 0.

After analyzing the relation between combine results and quarterback success metrics (Games Played, Touchdown/Interception Ratio, Completion Percentage), the evidence is clear, there is a weak correlation between a quarterback’s combine performance and his success as a professional football player. Out of the six NFL Combine measurements (40 Yard Dash Time, Vertical, Shuttle Time, Broad Jump, Weight, 3-Cone Drill Time), correlations only exist between a quarterback's weight and his completion percentage; making weight the only combine measurement that is predictive of NFL success at the quarterback position. Despite being a traditional and fun media event for fans, scouts, and players, based on the evidence from our sample of NFL quarterbacks from 2000-2020, the NFL Combine is not a good predictor of quarterback success.


Sources

https://www.pro-football-reference.com/

54 views

Recent Posts

See All