DiderotsGhost
Well-Known Member
- Joined
- Nov 28, 2011
- Messages
- 4,627
- Likes
- 23,503
I'm a pretty big fan of Daj2576's work here and I decided to do a similar analysis on Dan Mullen. For those unfamiliar with Daj, he has a model where he looks at recruiting results and then makes predictions based on that. The majority of coaches perform about in-line with expectations based on recruiting and long-run results suggest that most college football games can be predicted based on talent. Some coaches underperform talent levels, however, and some very good coaches outperform talent levels.
Based on my analysis, Dan Mullen has one of the most impressive track records of outperformance in the nation. He outperforms expectations by about 1.6 games per year. While there are other coaches that outperform by that margin, the reason I view Mullen's as the most impressive is that he's been doing it against elite competition. It's one thing to outperform by 1.6 games against MAC or Sunbelt competition. By and large, many of the programs in the mid-major conferences have pretty identical talent levels even if the recruiting rankings might vary somewhat (the #50 and #60 talent teams are pretty closely matched in reality), so a slight edge in coaching can make a huge difference in that environment.
The SEC West is a different beast altogether. There are 4 programs that routinely get elite top 10 classes (Alabama, LSU, Auburn, and Texas A&M) and Ole Miss has been up there as well in recent years. It's much more difficult to beat a top 10 talent team with a #30 team than it is to beat a #50 team with the #60 team. Yet, Dan Mullen consistently outperforms the talent level at Miss State. Moreover, Mullen has significantly improved Miss States recruiting performance over time. When you factor that into account, I estimate that Mullen has improved Miss States performance by about 2.3 wins per season.
The Model
My model is not as complex as Daj's in all likelihood. I worked for years in investment and needed to analyze things on a very quick basis. As a result, I often build "quick and dirty" models. Overall, my model is pretty simple. To analyze a team, I take the prior 5 years worth of recruiting data from 247 Sports. I then weight that data by class. Senior class counts for 40%, Juniors 30%, Sophomores 20%, Freshmen 5%, and Redshirt Seniors 5%. These percentages were derived from looking at about a half-dozen SEC rosters and seeing what percentage of starters and backups (2-deep) come from each class. My analysis found that Juniors and Seniors dominate rosters and Freshmen only rarely play. Thus, Freshman classes should be weighted low and Junior and Senior classes high.
Based on this methodology, here are the weighted average recruiting rankings for the 7 SEC West clubs since 2009.
First thing to note that it was annoying having 5 different teams in the SEC West that use red as one of their primary colors. I made Ole Miss blue, but that still left four more red teams. You can spot Miss State at the very bottom of the chart.
When Dan Mullen arrived at Miss State, it lagged behind the rest of the SEC West in talent by a very sizable amount. For the first few years at Miss State, Mullen was expected to lose every single game to an SEC West opponent. He has improved the talent level by about 11 spots in 9 years and if he were to stay at Miss State, the trend will continue to improve over the next few years. He currently has the #16 rated class for 2018 and would be expected to have a weighted average rating of about 23.0 in 2018. In talent terms, it would be the best team hes ever had.
My model is simpler than Dajs on wins and losses. I simply assign a W or L in most situations based on recruiting rankings. This isnt as precise as a probability-based model, but it should even out over the long-term in most cases for a team like Miss State (it wouldnt work as well for analyzing Alabama which has been #1 for several years in a row).
There are some exceptions. I decided to assign an effective tie in some spots (0.5 wins, 0.5 losses) when talent levels where relatively even and based on home-field advantage. For instance, if Miss State were rated #29 and played the #26 rated team at home, Id assign a half-win, half-loss. Likewise, if Miss State were rated #29 and played on the road against the #34 rated team, Id assign a half-win, half-loss, as well. However, the vast majority of games Miss State plays are either against teams much more talented or much less talented.
The Results
Here were the results of my model. Note that there are three charts here. The first is Expected Outcome based on recruiting ratings. The second is Actual Outcome. The third is Expected Outcome based on a hypothetical alternative reality where recruiting did not improve under Mullen. The first highlighted green row show Mullen's outperformance relative to his team's talent level; the second highlighted green row shows Mullen's outperformance relatively to the situation he inherited.
The takeaway is that Mullen has outperformed by a very wide margin. Over 9 seasons, hes outperformed his recruiting average by 14 games (1.6 games per year), and that will be 15 games (1.7 games per year) if he wins the Egg Bowl this week. However, even more notably, hes improved Miss States performance 20.5 games (21.5 with an Egg Bowl win) over 9 years based on a hypothetical alternative reality where recruiting does not improve. Depending on the Egg Bowl outcome, thats in the range of +2.3 to +2.4 wins per season that Miss State has improved over expectations based on Dan Mullens performance.
Mullen has outperformed 7 out of 9 years in my model and has only underperformed once (his first season, 2009). He outperformed by over 3 games 3 different times (2010, 2014, and 2017).
Ill remind you, these results came in the toughest division in all of college football. The results are also even more impressive than they look IMO. The reason for this is because the polarizing nature of Miss States schedule. By and large, its somewhat rare for Miss State to play games where they have comparable talent with the opposing team. Either they play teams with much greater talent or lesser talent, so its actually even more difficult than it appears to outperform there.
The chart below shows the weighted average recruiting rankings for all SEC teams.
Note how unusual it is to see a regular Miss State opponent within + or 5 spots. The only SEC teams with comparable talent on Miss States schedule have been Arkansas (2012 present), Kentucky (2016 present), and few seasons of Ole Miss (2012 2015). Miss State has been at a severe disadvantage against Alabama, LSU, Auburn, and Texas A&M for the entire timeframe under Mullen (note that A&M didnt start competing in the SEC till 2012.) Moreover, due to Ole Misss shady recruiting tactics, Miss State has been playing the Egg Bowl at a significant disadvantage, as well. Add in the occasional cross-division game against Florida, Georgia, Tennessee, or South Carolina, and Miss State regularly plays 5 6 games per season at a severe talent disadvantage.
That is what makes Mullens performance truly remarkable. Hes outperforming by a wide margin against a slate that is among the most difficult to outperform against in college football.
Butch Jones vs Dan Mullen in the Model
For those wondering how Butch Jones fared in my model, surprisingly he was about even until this season. My model shows underperformance by 2 games in 2013 (which shouldve been expected given the situation he inherited was much worse than represented by the recruiting rankings) and -1 in 2014 (also forgivable). He outperformed by +0.5 games in 2015 and 2016, in spite of the perception that he underperformed (thank Josh Dobbs). In 2017, however, my model expects Tennessee to go 9-3 and we know that the actual performance will end up being either 5-7 or 4-8. So Butch basically underperformed by somewhere between 4 and 5 games this season; a massive underperformance! It's particularly noteable that Florida's weighted average rating fell significantly while ours climbed significantly, so that we had more talent than Florida for the first time in over a decade.
If we assume that Mullen had taken over 2013 and he had a similar outperformance at Tennessee as he did at Miss State, then based upon our talent levels, we might have expected him to achieve results somewhat in line with this:
2014: 9-4
2015: 10-3
2016: 10-3
2017: 10-2
This is merely speculation, of course, but even if Mullen merely met expectations with the 2017 talent, he would be 9-3 this season.
Dan Mullen would be a Top Tier Hire
This is the mathematical case for why I view Dan Mullen as a top tier hire. I'd add that there's also a very strong qualitative case given that he's one of the best developers of Quarterbacks in the nation (Alex Smith, Tim Tebow, Dak Prescott, Nick Fitzgerald) and he won 2 national titles at Florida as OC. Id go so far as to suggest that Mullen's in the debate for Best College Football Coach in America.
Of course, its nevertheless difficult to evaluate how Mullen would perform at a top-tier program. Hes improved recruiting at Miss State significantly, but Tennessee was already recruiting well under Butch Jones, so its not necessarily automatic that Mullen would improve recruiting further. That said, its certainly possible.
Likewise, while Dan Mullens record against the top 20 seems weak, remember that based on talent, he should have somewhere close to a 0% winning percentage against the top 20 teams hes faced. Its tough to know how he would fare against Nick Saban and Kirby Smart with top 5 talent. I think the evidence says hed win a good percentage of the time, but well never know until we see him with elite talent.
Regardless to me, Dan Mullen is a low-risk hire with high upside potential. Hes a coach that appears to be good enough to win a national title (and dont forget that he won 2 of them at Florida). Is he the #1 coach available? Maybe. I tend to favor Scott Frost slightly if we can get him, but I view Dan Mullen as the most underrated coach in all of college football right now. The guy is a winner. He took a job that could be best described as a coaches graveyard and has significantly exceeded expectations. If he can do the same at Tennessee, I think wed make the Playoff at least twice in the next decade.
I don't think there's any coach available that is a "sure-fire success" at Tennessee, but Mullen certainly seems like the "least likely to fail" who also has the potential to compete for a national title.
Based on my analysis, Dan Mullen has one of the most impressive track records of outperformance in the nation. He outperforms expectations by about 1.6 games per year. While there are other coaches that outperform by that margin, the reason I view Mullen's as the most impressive is that he's been doing it against elite competition. It's one thing to outperform by 1.6 games against MAC or Sunbelt competition. By and large, many of the programs in the mid-major conferences have pretty identical talent levels even if the recruiting rankings might vary somewhat (the #50 and #60 talent teams are pretty closely matched in reality), so a slight edge in coaching can make a huge difference in that environment.
The SEC West is a different beast altogether. There are 4 programs that routinely get elite top 10 classes (Alabama, LSU, Auburn, and Texas A&M) and Ole Miss has been up there as well in recent years. It's much more difficult to beat a top 10 talent team with a #30 team than it is to beat a #50 team with the #60 team. Yet, Dan Mullen consistently outperforms the talent level at Miss State. Moreover, Mullen has significantly improved Miss States recruiting performance over time. When you factor that into account, I estimate that Mullen has improved Miss States performance by about 2.3 wins per season.
The Model
My model is not as complex as Daj's in all likelihood. I worked for years in investment and needed to analyze things on a very quick basis. As a result, I often build "quick and dirty" models. Overall, my model is pretty simple. To analyze a team, I take the prior 5 years worth of recruiting data from 247 Sports. I then weight that data by class. Senior class counts for 40%, Juniors 30%, Sophomores 20%, Freshmen 5%, and Redshirt Seniors 5%. These percentages were derived from looking at about a half-dozen SEC rosters and seeing what percentage of starters and backups (2-deep) come from each class. My analysis found that Juniors and Seniors dominate rosters and Freshmen only rarely play. Thus, Freshman classes should be weighted low and Junior and Senior classes high.
Based on this methodology, here are the weighted average recruiting rankings for the 7 SEC West clubs since 2009.
First thing to note that it was annoying having 5 different teams in the SEC West that use red as one of their primary colors. I made Ole Miss blue, but that still left four more red teams. You can spot Miss State at the very bottom of the chart.
When Dan Mullen arrived at Miss State, it lagged behind the rest of the SEC West in talent by a very sizable amount. For the first few years at Miss State, Mullen was expected to lose every single game to an SEC West opponent. He has improved the talent level by about 11 spots in 9 years and if he were to stay at Miss State, the trend will continue to improve over the next few years. He currently has the #16 rated class for 2018 and would be expected to have a weighted average rating of about 23.0 in 2018. In talent terms, it would be the best team hes ever had.
My model is simpler than Dajs on wins and losses. I simply assign a W or L in most situations based on recruiting rankings. This isnt as precise as a probability-based model, but it should even out over the long-term in most cases for a team like Miss State (it wouldnt work as well for analyzing Alabama which has been #1 for several years in a row).
There are some exceptions. I decided to assign an effective tie in some spots (0.5 wins, 0.5 losses) when talent levels where relatively even and based on home-field advantage. For instance, if Miss State were rated #29 and played the #26 rated team at home, Id assign a half-win, half-loss. Likewise, if Miss State were rated #29 and played on the road against the #34 rated team, Id assign a half-win, half-loss, as well. However, the vast majority of games Miss State plays are either against teams much more talented or much less talented.
The Results
Here were the results of my model. Note that there are three charts here. The first is Expected Outcome based on recruiting ratings. The second is Actual Outcome. The third is Expected Outcome based on a hypothetical alternative reality where recruiting did not improve under Mullen. The first highlighted green row show Mullen's outperformance relative to his team's talent level; the second highlighted green row shows Mullen's outperformance relatively to the situation he inherited.
The takeaway is that Mullen has outperformed by a very wide margin. Over 9 seasons, hes outperformed his recruiting average by 14 games (1.6 games per year), and that will be 15 games (1.7 games per year) if he wins the Egg Bowl this week. However, even more notably, hes improved Miss States performance 20.5 games (21.5 with an Egg Bowl win) over 9 years based on a hypothetical alternative reality where recruiting does not improve. Depending on the Egg Bowl outcome, thats in the range of +2.3 to +2.4 wins per season that Miss State has improved over expectations based on Dan Mullens performance.
Mullen has outperformed 7 out of 9 years in my model and has only underperformed once (his first season, 2009). He outperformed by over 3 games 3 different times (2010, 2014, and 2017).
Ill remind you, these results came in the toughest division in all of college football. The results are also even more impressive than they look IMO. The reason for this is because the polarizing nature of Miss States schedule. By and large, its somewhat rare for Miss State to play games where they have comparable talent with the opposing team. Either they play teams with much greater talent or lesser talent, so its actually even more difficult than it appears to outperform there.
The chart below shows the weighted average recruiting rankings for all SEC teams.
Note how unusual it is to see a regular Miss State opponent within + or 5 spots. The only SEC teams with comparable talent on Miss States schedule have been Arkansas (2012 present), Kentucky (2016 present), and few seasons of Ole Miss (2012 2015). Miss State has been at a severe disadvantage against Alabama, LSU, Auburn, and Texas A&M for the entire timeframe under Mullen (note that A&M didnt start competing in the SEC till 2012.) Moreover, due to Ole Misss shady recruiting tactics, Miss State has been playing the Egg Bowl at a significant disadvantage, as well. Add in the occasional cross-division game against Florida, Georgia, Tennessee, or South Carolina, and Miss State regularly plays 5 6 games per season at a severe talent disadvantage.
That is what makes Mullens performance truly remarkable. Hes outperforming by a wide margin against a slate that is among the most difficult to outperform against in college football.
Butch Jones vs Dan Mullen in the Model
For those wondering how Butch Jones fared in my model, surprisingly he was about even until this season. My model shows underperformance by 2 games in 2013 (which shouldve been expected given the situation he inherited was much worse than represented by the recruiting rankings) and -1 in 2014 (also forgivable). He outperformed by +0.5 games in 2015 and 2016, in spite of the perception that he underperformed (thank Josh Dobbs). In 2017, however, my model expects Tennessee to go 9-3 and we know that the actual performance will end up being either 5-7 or 4-8. So Butch basically underperformed by somewhere between 4 and 5 games this season; a massive underperformance! It's particularly noteable that Florida's weighted average rating fell significantly while ours climbed significantly, so that we had more talent than Florida for the first time in over a decade.
If we assume that Mullen had taken over 2013 and he had a similar outperformance at Tennessee as he did at Miss State, then based upon our talent levels, we might have expected him to achieve results somewhat in line with this:
2014: 9-4
2015: 10-3
2016: 10-3
2017: 10-2
This is merely speculation, of course, but even if Mullen merely met expectations with the 2017 talent, he would be 9-3 this season.
Dan Mullen would be a Top Tier Hire
This is the mathematical case for why I view Dan Mullen as a top tier hire. I'd add that there's also a very strong qualitative case given that he's one of the best developers of Quarterbacks in the nation (Alex Smith, Tim Tebow, Dak Prescott, Nick Fitzgerald) and he won 2 national titles at Florida as OC. Id go so far as to suggest that Mullen's in the debate for Best College Football Coach in America.
Of course, its nevertheless difficult to evaluate how Mullen would perform at a top-tier program. Hes improved recruiting at Miss State significantly, but Tennessee was already recruiting well under Butch Jones, so its not necessarily automatic that Mullen would improve recruiting further. That said, its certainly possible.
Likewise, while Dan Mullens record against the top 20 seems weak, remember that based on talent, he should have somewhere close to a 0% winning percentage against the top 20 teams hes faced. Its tough to know how he would fare against Nick Saban and Kirby Smart with top 5 talent. I think the evidence says hed win a good percentage of the time, but well never know until we see him with elite talent.
Regardless to me, Dan Mullen is a low-risk hire with high upside potential. Hes a coach that appears to be good enough to win a national title (and dont forget that he won 2 of them at Florida). Is he the #1 coach available? Maybe. I tend to favor Scott Frost slightly if we can get him, but I view Dan Mullen as the most underrated coach in all of college football right now. The guy is a winner. He took a job that could be best described as a coaches graveyard and has significantly exceeded expectations. If he can do the same at Tennessee, I think wed make the Playoff at least twice in the next decade.
I don't think there's any coach available that is a "sure-fire success" at Tennessee, but Mullen certainly seems like the "least likely to fail" who also has the potential to compete for a national title.
Last edited:
