Current NFL Head Coach Relationship Analysis

Project Proposal

Every athlete in the world has once dreamt of playing the sports they love professionally; I was not different by any means. Despite learning what “gridiron football” is when I moved to Virginia prior to my freshman year of high school, I fell in love with the sports and hoped to play as long as possible. Summer leading into my senior year, our head coach changed. We changed our offensive scheme from the “Wing-T” offense to a “Pro-Style Offense.” “Wing-T” is a more traditional and old-school type of offense that relies on proper execution of a small number of plays. The–supposed–strength of it allows smaller, less physically talented players to win through out-executing and being disciplined; that is the reason why service academies run “option” offense, which is an evolved version of “Wing-T” offense. On the other hand, the “Pro-style” offense was more versatile: more formations, more plays, more variations. I had to understand a completely different set of football information and I loved it! That’s when I knew that I truly loved football not only for the excitement of playing football but also for the Xs and Ox behind football. Coming from a less-athletically competitive school, I knew that the chance of me even playing football at the collegiate level was slim. However, I was fortunate enough to receive an opportunity to prolong my athletic career at Washington and Lee University. Although I cannot play at a bigger Division I school and aim to be drafted to the National Football League, my desire to become a football coach in the NFL is bigger than ever. This project looks into the network of current NFL head coaches and some major assistant coaches to visualize the underlying hiring practice that is prevalent in the coaching realm at large. As a person who dreams of being in the NFL one day as a coach, learning how hiring practice works in the National Football League will greatly benefit what I need to work on in order to be a coach one day.

A couple of years ago, I watched a YouTube video that illustrated the impressive coaching tree of Andy Reid, current Head Coach of Kansas City Chiefs and most recent Super Bowl champion. Although 2019 was his first year of coaching the Super Bowl team, his former assistant has had success in the NFL as Head Coaches. The success of his assistant coaches was quite different than that of Bill Belichick’s assistant. Belichick, who is often considered one of the greatest coaches of the NFL, produced numbers of head coaches, yet they seemed to never pan out as well as that of Reid’s. It would be interesting to visually compare the two major coaching tree in the NFL.

I am going to look at the most recent coach roster for teams and proceed in reverse-chronological order. Considering I am trying to incorporate as many teams as possible, the source of my data will not be limited to a single franchise, rather I would have to personally gather necessary information on both franchise, year, head coaches, assistant coaches, etc. I have found some outlets that I can collect data from: (1)https://www.profootballarchives.com/nfl-head-coaches.html and (2) https://www.pro-football-reference.com/coaches/. I have found two studies that have specifically looked into the networks of coaching–so-called “Coaching Trees” in the NFL. Firstly, The NFL Coaching Network: Analysis of the Social Network Among Professional Football Coaches by Andrew Fast and David Jensen gave me inspiration and fundamental background information on modeling how I am going to construct the basis of my project. Secondly, An analysis of National Football League coaching trees and the network they comprise by McCullick et al. pointed me to what and where I can find the relevant data.

I am not entirely sure of the method of analysis. I will have a more concrete idea on how I am going to dissect the vast data as I work on the project.

As aforementioned, I am planning to visually disclose the hiring practices at the National Football League. Further data gathering on performance and hiring can be used to answer a question such as whether the performance or the connection is more important in getting hired as a coach for a franchise in the NFL. It is hard to categorize the exact discipline of this research. However, I believe this can be applied to various different sports not limited to football as well as hiring practice in different industries at large.

Success for my project would look like an interactive network visualization app that allows one to see not only the relationship but also personal information.

Success for my long-term goal and career will be starting my own coaching and putting myself on my own coaching tree visualization app!

Project Update (10/23/2020)

The data collection went easier than I thought it would. Before actually collecting the data, I did not take into account of the coaching career at the college level; some coaches only coached college before recently being an NFL coach, some coaches go back and forth between NCAA and professional level. The Pro Football Reference website nicely organized the coaching career of every coach; having all the information that I am interested in on one website was definitely helpful and efficient.
The tough part regarding the project so far was the formating of the collected data. Since various different programs utilized different algorithms to create a visualization of a network, they all utilize unique data format. Gephi that Professor Brooks had a unique way of formating data files. We shall see if the altered format for my data will fit the program and properly function.
The lesson of the week for me was that I learned how important adaptability is. Although Gephi is a great tool, as a person who has only used Palladio, the Node and Edge table were unique and somewhat hard to understand/adapt. My initial thinking after the reading tutorial was “This is too difficult and hard to understand.” Although I have great resources for asking Professor Brooks, not everyone has that kind of luxury and will probably give up on using it. Therefore, making it user friendly is very important. This can be utilized if I want to develop a football statistic analytics tool, I must make it user friendly.
My goal next week is finishing data collection in proper format and testing if the format works on Gephi!

Project Update (10/31/2020)

The overall feedback that I received was very positive. I was initially grouped with Gabriel and Sunghee who also did network analysis. Although my project was focused on the current NFL head coaches, Sunghee and Gabriel both understood the main topic with few of the notable names such as Bill Belichick, Kyle Shanahan, etc. However, they pointed out that to a stranger who knows nothing about football, these names will be unfamiliar. Hence, in my description/caption of the first visualization, I made sure to include the fact that this visualization is the network of current NFL head coaches and what does having a connection means, and what the thickness of the line means.

Sunghee used Palladio which allowed her to differ the size of the nodes on her visualization. I think that would be cool/help distinguish the teams of the coaches. I was also grouped with Jack Jones and Ryan Brands, who did different visualization than me. Looking at Jack’s project I learned to correctly label my visualization because he mislabeled it twice!

I was quite surprised by how much people liked the colors. Since I had to denote at most 32 teams/nodes, I had very limited choices in the color pallet.

I am not sure how to put labels on the Gephi itself; hence, I might make a separate box that denotes every coach with their individual team. Maybe I will also upload the screenshot of the node table and key for team names to further support the audience’s comprehension of my visualization.

I am not completely sure what I can do to further improve my data! I may look into whether their parents/other family members worked in the football industry. I also should look into hometowns and DOB to see who is the oldest/youngest.

Results

I vividly remember the moment that I started to enjoy Xs-and-Os of football. However, wanting to become a football coach came much later because I was unsure; I was unconfident whether I could get myself into such a tight-knit community. I always surmised that the NFL coaches are a selective group of individuals: everyone has worked with each other, and hiring occurs within that small network of coaches. As an individual who strives to pursue a career in coaching football, the analysis of current NFL head coaches will not only confirm my previous belief but also provide me with a potential direction to approach the employment.

Through this project, I have learned that the head coaches of the National Football League are indeed tightly intertwined group of individuals who have previously worked together or you have worked with one’s coworkers. Such nature of hiring practice suggests that the NFL franchise often finds it necessary for a coach to have years of experience working with the professional athlete at the highest level of competition. Furthermore, it illustrates the somewhat closed-minded hiring practice; nevertheless, such hiring practice suggests that once an individual makes it to an entry-level coaching job at the NFL, the chance of promotion and getting a head coaching job increases.

One interesting aspect of football is each organization and each individual differs in their role in the game time even though they might have the same title. However, the general public usually associates the success of a unit–offense or defensive–with the appropriate coordinator. When a franchise hires an offensive or a defensive coordinator of a successful team, the new franchise rightly expects the new head coach to produce similar results. However, based on the analysis of current NFL head coaches, that wasn’t always the case. For example, Bill Belichick, a current NFL head coach who is called one of the greatest–if not the greatest– coach of all time have produced numbers of head coaches from his assistance and co-workers (Romeo Crennel, Matt Patricia, Joe Judge, Brian Flores), yet none of them has a winning record greater than or equal to .500. On the other hand, the majority of Andy Reid’s assistant coaches have flourished in the NFL as a head coach. The bipolar records of assistant coaches of two great coaches were quite shocking and prove that not all great head coaches (mentor) can engender great assistants (mentee). Furthermore, some of the coaches, who have coached in the NFL for a long time, did not have any of their former co-coaches as a current head coach. Pete Carroll, a longtime NFL coach and Super Bowl-winning-coach, does not have any of his former co-coaches as a head coach.

The NFL head coach network analysis result can be looked at from various angles and disciplines. Firstly, the result can be interpreted from organizational behavior discipline. Although this project is limited to the National Football League hiring practice, it can be modified and adapted into various professions, especially in which employees are highly specialized. For example, various studies have been done especially, on, promoting someone “in house” rather than experimenting with outsiders. Secondly, this result can be assessed by the sports science discipline. Various sports science disciplines have investigated what leads to great coaching/winning. The result of this project showcases that a great mentor with winning records do not always engender great mentees. Further research can be done on the difference between Andy Reid and Bill Belichick to determine what are the differences between two head coaches.

I believe the next step to this project is to look at all the head coaches in NFL history or at least the modern Super Bowl era. I did not have the time nor ability to record all the information about all the head coaches in NFL history. A step further would involve visualization of all the coaches in NFL history, not limited to the head coaches. It would be interesting to see how coach-hiring practices differ from the collegiate level to the professional NFL level. On a side note, the next step on the project on my end would be finding an entry-level coaching job at the NFL level and work my way up the ladder.

The biggest challenge with my method is that my method does not take into account the impact of non-accounted factors. Hence, the network between coaches is very limited. For example, I did not account for any family relationship nor a player-coach relationship. For example, Mike LaFleur, an assistant to Kyle Shanahan is a brother of Matt LaFleur, a Green Bay Packers Head Coach. Furthermore, Mike Vrabel, a current Tenessee Titans Head Coach, served as a player under Bill Belichick while he was a player in the NFL. As aforementioned, the number of individuals is limited to 32 at the moment, failing to showcase the vast amount of relationship between current and former assistant coaches. In terms of the problem with visualization, the current method of visualization is static: it captures the network with a specific filter. The fact that the audience cannot manipulate the filters limits the extent of visualization. If I had more HTML or any coding skill, I would’ve love to create an interactive version of this visualization to allow the audience to select and change the filters.

The network of 32 Current NFL Head Coaches: an individual line indicates having coached together previously; the thickness of a line indicates the frequency of co-coaching. As you can see most NFL head coaches have worked with each other at some point in their coaching career. Therefore, it is more likely to land an head coaching job if you have worked in an NFL team with other HC’s.

.500 or Above Winning % (Top) & Below .500 Winnning %(Below) The two visualization go hand in hand with each other. Although Bill Belichick has lots of connections, amongst his connections, Belichick himself is the only coach with a 500 or better winning percentages. On the other hand, Andy Reid’s assistant/co-coaches are doing well in the NFL.