Our project analyzed the effectiveness of the NFL combine in predicting the career readiness of offensive players in the NFL. Specifically, we focused on the offensive players who participated in the combine in 2018 and examined their first-year success in the league. By examining the correlation between combine performance and NFL career success, we aimed to provide insights to improve how NFL teams evaluate players and guide athletes in their preparation for their professional careers.
The NFL combine holds significant weight in an NFL hopeful's draft stock, often influencing team decisions and players' training priorities. However, there have been notable cases, such as Tom Brady, where poor combine performance did not hinder an athlete's remarkable success in the league. Therefore, we aimed to determine if the combine accurately identified NFL-ready players across different positions or if its importance should be reassessed.
We hypothesized that the degree to which combine performance predicted NFL readiness would vary by position. For example, we expected that positions requiring more skill, like quarterbacks, may not be as reliant on combine scores as other positions like linebackers. Furthermore, we anticipated that even in positions where combine performance was a strong predictor of NFL success, there would be exceptions—athletes with exceptional combine performances who did not excel in the league.
Ultimately, our analysis shed light on the benefits and limitations of the NFL combine, helping teams make more informed decisions when evaluating players and aiding athletes in understanding the most effective areas to focus their training efforts. By evaluating the relationship between combine performance and NFL career success, we contributed to the improvement of player analysis and preparation strategies in the NFL.
The NFL Raw Data dataset was downloaded from the Advanced Sports Analytics website. It is a CSV file containing player fantasy football data and game information for each game from 2019 to 2023. The dataset had 26,600 rows and 68 columns, with a file size of approximately 6.7 MB. The structure of the dataset consisted of multiple rows per game, with each row corresponding to a player's performance in that game. The columns of interest were "winning_team," "vis_team," and "home_team." These columns were used to determine the winning team for each game. This dataset was used to calculate the win/loss percentage of each player's team in their first year and make adjustments to their career success scores based on this ratio.
The NFL Combine Data was obtained by web scraping from the NFL Combine Results website. The data was retrieved for the year 2018 and specifically for quarterbacks. The data was organized in an HTML table format. The structure of the dataset included rows corresponding to players, with columns representing different combine metrics such as "Name," "40 Yard Dash," "Vertical Leap," "Broad Jump," "Shuttle," and "3Cone." These metrics were used to calculate the combine score for each player. The dataset provided insights into the combine performance of players, which was then combined with other data to analyze the correlation between combine performance and NFL career success.
The ESPN Sport Core API Data was used to retrieve NFL player information, including career statistic and college career data. The data was obtained by making API requests to the ESPN API. The structure of the dataset was in JSON format and contained nested dictionaries. Each player's data was stored under the "athlete" key, with nested dictionaries representing different seasons of their career. This dataset provided comprehensive information about player performance across their entire NFL career. It was used to calculate career success scores and gain insights into the relationship between player performance and career success. The data from this dataset was combined with other datasets, such as combine scores, to draw meaningful conclusions.
In the data cleaning process, several steps were taken for each dataset to ensure the accuracy and consistency of the data.
For the Combine Data, empty rows were removed to eliminate any unnecessary data. Empty strings were converted to "None" values to differentiate them from actual data entries. Additionally, position abbreviations in the dataset were standardized to match the abbreviations used in the API data.
In the Game Log Data, missing team names in the team name columns were filled by utilizing context clues from other available information. This involved examining the game details and cross-referencing with team information to assign the correct team names. Furthermore, a new column called "winning_team" was added to indicate the team that won each game. This was determined by aggregating the total number of wins per team.
For the API data, a pandas dataframe was created to organize the statistics that varied based on the player's position. This allowed for efficient data manipulation and analysis of the player statistics.
These data cleaning processes were crucial in ensuring the integrity and consistency of the data across the datasets. By addressing missing values, standardizing formats, and aggregating relevant information, the datasets were prepared for further analysis and insights to be drawn.
Our analysis revealed a strong positive correlation between combine performance and career success for running backs (RBs) and wide receivers (WRs). This suggests that RB and WR prospects should prioritize their preparation for the NFL combine as it serves as a reliable indicator of their potential success in the league. However, an interesting finding was a negative correlation between combine performance and career success for safeties (Ss), implying that standout combine performances may not necessarily translate to success for this position.
Among the top combine performers in 2018, only Josh Allen stood out as a successful NFL player among quarterbacks (QBs). Nick Chubb was the sole running back (RB) who excelled at the combine and went on to have a good NFL career. Surprisingly, none of the wide receivers (WRs) or safeties (Ss) who dominated the combine have become prominent names in the NFL. This insight highlights the importance of considering on-field performance and other factors alongside combine results when evaluating prospects.
Analyzing touchdown scorers since 2019, we found that while Saquon Barkley received more attention for his combine performance, Nick Chubb has proven to be a slightly more effective NFL player. This analysis demonstrates the value of identifying productive players who may be flying under the radar and could be high-value acquisitions. Jeff Wilson Jr., for example, emerged as one of the most productive players from the 2018 combine, showcasing the potential of finding hidden gems in player analysis.
Athletes who exhibited exceptional athleticism with a sub-4.5 40-yard dash time and a minimum of 20 reps on the bench press were considered athletic anomalies. While this combination of speed and strength is theoretically advantageous, few athletes in this category have achieved significant success in the NFL. Saquon Barkley stands out as an exception. Notably, Shaquem Griffin, despite having only one hand, excelled at the 2018 combine, demonstrating his determination and remarkable abilities.
The variance of 40-yard dash times differed across positions in the 2018 combine. Running backs (RBs) and safeties (Ss) displayed remarkably low variance, suggesting that focusing on improving their 40-yard dash times can significantly differentiate them from other athletes in their positions. On the other hand, linebackers (LBs) and quarterbacks (QBs) exhibited the highest variance, indicating that speed may be less crucial for these positions, or there is a wider range of talent levels competing in the combine for these positions.
This visualization presents a correlation matrix, depicting the relationships between combine scores and career success for each position. The intensity of the color indicates the strength of the correlation, with deeper red representing a stronger positive correlation and blue indicating a negative correlation. This visually confirms our first insight, highlighting the strong positive correlation between combine performance and NFL success for running backs (RBs) and wide receivers (WRs), while revealing the unexpected negative correlation for safeties (Ss).
In this interactive visualization, you can explore the top five combine performers for each position in the 2018 NFL combine. By selecting a position from the dropdown menu, the bar plot dynamically updates to display the names and career scores of the top performers for that position. This visualization aligns with our second insight, emphasizing the limited NFL success of most top combine performers across positions, with only a few exceptions such as Josh Allen and Nick Chubb.
This interactive visualization presents box plots for various combine drills, allowing you to select and examine any specific drill of interest. The vertical leap drill, for example, measures explosiveness. The box plots for wide receivers (WRs), running backs (RBs), and safeties (Ss) reveal similar median and 75th percentile scores, but Ss show a significantly tighter distribution. On the other hand, linebackers (LBs) and WRs exhibit wider ranges, indicating opportunities for standout performances in explosiveness. This visualization supports our third insight, highlighting the varying distributions of combine scores across positions.
This interactive visual illustrates the top x percent of touchdown leaders among players who debuted in 2018, where x is determined by a slider. Please note that the displayed statistics only reflect the players' 2022 performance, as a discrepancy was discovered in the dataset. Nonetheless, this visualization aligns with our third insight, showcasing the productive players from the 2018 combine who have excelled in scoring touchdowns. It offers a glimpse into the potential hidden gems and high-value acquisitions among this group of players.
Our project features a Graphical User Interface (GUI) that provides users with comprehensive career score data for NFL players who debuted in 2018. The GUI allows users to explore player statistics such as receptions, receiving yards, receiving touchdowns, solo tackles, forced fumbles, interceptions, passes defended, rush yards, rush touchdowns, and sacks. Users can easily navigate and sort the data by searching for a specific player's name using the search bar or selecting a position from the drop-down menu. The GUI offers a user-friendly interface for seamless data exploration and analysis. To access the code for the GUI, refer to the "PhaseIII.py" file in the GUI folder.