We are thrilled to welcome a very special guest this week to our “4 Questions with the Coach” feature. Loren Maxwell is a freelance sports analyst and a columnist with the Atlanta Journal- Constitution, and you have more than likely seen and read his statistical breakdowns, projections, and rankings, known as the Maxwell Ratings, in the AJC and in Todd Holcomb’s newsletter, the Georgia High School Football Daily. Loren is the creator of the Maxwell Ratings, which include preseason projections, as well as weekly projections; both appear in the AJC and GHSF Daily.
Loren also is the founder of the Georgia High School Football Historians Association, which offers historical data and information for games, players and teams in the state. He now manages similar historical sites dedicated to Arkansas, South Carolina, and Tennessee high school football.
Loren’s educational background includes an MBA from the University of Cincinnati and an MPA from Bowie State University. Let’s hear more about how Loren uses the magic of statistical analysis and historical data to project how many games your favorite team could win this year, or if your team has any chance of winning its region and making the playoffs.
Q. Loren, thanks so much for agreeing to join us for this week’s 4 Questions segment. While most people reading this have likely seen your Maxwell Ratings, can you tell us exactly what it means and what can the average fan take from the ratings?
A. The Maxwell Ratings are the best mathematical explanation for what we see on the field after taking into account each team’s historical performance and their current season performance. The mathematical basis of the model, known as maximum likelihood, is a common approach taken in many other industries, such as the medical industry, business operations, etc., and I’ve only adopted it for this purpose.
Each team is a variable in the model with a particular value (their rating), and each game is an equation in the model with a result (Team A score – Team B score = margin of victory). It’s possible to determine how likely that result is, given the ratings of the teams.
As games are played, the program calculates the likelihood of each outcome based on the current set of ratings, totals that likelihood for the entire season, and then adjusts the rating of each team up or down to increase that total likelihood. Once none of the ratings can be adjusted up or down to improve the total likelihood for the season, the maximum likelihood has been found and the ratings are finalized. This represents the best mathematical explanation for the season.
This can sometimes lead to counterintuitive results since many fans focus on the outcome of a specific game or a small set of games, but the program has the advantage of considering all games at once when determining the rating for each team. However, a football game has many random events, so while the ratings may say a particular outcome is most likely, of course many other outcomes could occur.
Q. Have you always had a love and interest in high school football, and how did the Maxwell Ratings idea originate exactly?
A. Actually my main interest has always been college football, which I started to follow in the 1980s when I was a teenager. At that time I’d get USA Today and follow Jeff Sagarin’s college football ratings. However, I grew up in Athens and went to Clarke Central High School, and in the 1980s we had one of the premier football programs in the state, if not the nation. At the time the Atlanta Journal-Constitution published all the scores and schedules of the high school football teams on Sunday, so the data was easily available, and so I started to play around with different ideas to emulate Sagarin’s ratings for Georgia high school football.
Today I’d be embarrassed by those early attempts at the ratings, but as I tried different things and got more educated in math, I eventually took several large steps toward the model I use now, which I would argue based on the college football ratings is superior to Sagarin’s, my childhood idol!
Q. Let me switch over to the Georgia High School Football Historians Association. How did that come about?
A. I did a 22-year career in the Air Force and have lived in several different statesand in Europe. That’s mainly why my interest in Georgia high school football has varied, because most of that time was before the internet and I could not easily access information about what was going on in the state. I remember being surprised to find out Clarke Central had played in the 1992 state championship game about a year after it happened.
However by November 2003, the internet had started to make it somewhat possible to track current events and to also connect with others who had similar interests. So while I was stationed for the Air Force in New Jersey, I reached out to around 30 folks who were reporters or had websites centered on Georgia high school football and suggested creating the Georgia High School Football Historians Association.
The response was largely positive and we built a good collection of people to help get the organization off the ground. I cannot possibly present a comprehensive list of the many valuable contributors, but Bobby Hodges built our first website, which was instrumental in really establishing our presence.
Early on Todd Holcomb of the AJC did a large amount of the initial score entry, along with Becky Taylor, Steve Slay, Don Fendley, and Ira Kirkpatrick doing a huge amount of research on everything from the polls to the coaches to all-star games, etc.
Due to several deployments in the 2000s, I became largely uninvolved with the organization, and it’s a testament to the other volunteers that they were able to not just keep everything running, but make a huge amount of progress filling in the history of the state.
After I retired from active duty in 2012, Bobby turned the site over to me and I have been managing it ever since, to include adding Tennessee from Josh Easter’s former site, South Carolina from Dave Pickren’s former site, and Arkansas with the help of Rich Sanders. I also have plans to eventually add more states with the help of David Parker of the Alabama and Mississippi sites and anyone else who might step forward to assist.
Q. The Maxwell Ratings is all about the data, and I love it, as do most fans of Georgia high school football. However, I have always wanted to ask you about the fact that a team’s personnel, as in players lost to graduation, returning starters,etc., do not factor in at all in your preseason rankings. I find that interesting. Is that simply a factor not measurable in your current methodology? Have you ever considered finding a way to perhaps include such critical information in future rankings?
A. Factoring in a team’s personnel presents some significant challenges, not the least of which is the difficulty in getting that data, if it even exists, especially for lower-classification teams.
Additionally, if I had that data and were to factor in returning starters, then there would be additional questions to resolve, such as how I should treat a season-ending injury to a returning starter once the season begins or whether a backup is as good as the starter or whether the bench is two or three deep in certain positions, etc.
I think those questions at the high school level are largely unanswerable because of the lack of data, and even if the data were available and were accounted for, I’m not convinced it would tell us significantly more than the historical performance of the team itself already does.
I could see a much stronger argument for factoring in coaches, which is easier to determine, and for many there would even be some historical data, but again, I’m not sure the return on the effort would be worth it above just using a team’s historical performance. However, the preseason ratings diminish fairly quickly in the model anyway, so that essentially by the fourth game its impact is almost non-existent.
