Once upon a time I did an iPhone app showing the UNC 2-deep depth chart. One of the app's features was the designation of recruiting studs from players weren't. My "stud" criteria was a bit arbitrary; Top 10 in NC or equivalent in another state. (I have an elaborate definition of what "equivalent" means, but it is essentially 4 and 5-stars).
Since 2016 I've researched 37 teams' 2-deeps, calculated percentages of studs on those 2-deeps, and calculated the average academic classes for the 2-deeps. I combined those figures, as well as the coach's all-time record, into a multiple regression analysis to estimate winning percentage and found some interesting conclusions.
Correlation Factors
(0.0=none, 1.0=absolute)
- Overall Experience (-0.04) - Experience has very little impact on the Win/Loss record; if any, it is negative.
- Offensive Experience (-0.019) - Almost no correlation
- Defensive Experience (-0.103) - This is significant, and negatively correlated. Remember that when people praise defensive experience.
- Overall Stud Factor (0.491) - It's about the Jimmys and Joes. Recruiting matters
- Offensive Stud Factor (0.405) - Offense matters, but ...
- Defensive Stud Factor (0.543) - Defensive studs are the most important player-related factor leading to the Win/Loss record.
- Coach's Overall W/L Record (0.459) - Coaching matters. Of the coaches in this small study, those who didn't win more than 60% of their career games didn't stick around long. Records ranged from 0.329 (Derek Mason as of 2016) to 0.803 (Nick Saban as of 2016). Mack Brown, incidentally, is 0.659, a bit higher than the average record in the study (0.602).
The Equation
Finding an equation to predict a record is a matter of trying the most correlative factors against the historic W/L records. We can get a correlation factor (R-Squared) of 0.938 if we lump 8 factors in, but it is a cluttered mess. In fact we can settle for an R-Squared of
0.924 by going with just two factors. Here is the equation:
WinLossPct = 0.001818*STUD + 0.00827064*COACH
STUD = The percentage of the overall 2-deep that is studs
COACH = The all-time Win/Loss percentage of the coach.
This UNC team has
44% of its 2-deep filled with studs (=0.080011)
Mack Brown has won
65.9% of his games (=0.545)
Add them together and you get
0.625. Multiply that by 12 and you get
7.5 wins.
Given the number of studs on our defense, the introduction of a decent defensive coordinator, and the poor play of the ACC, I think this team will likely be an 8-win team, possibly 9.
Of course injuries are the big wildcard. However the lack of talent and propensity for injury in the UNC secondary will be a weakness that opponents exploit. UNC's other weakness is the talent level and high age of the offensive line. Couple that with the pattern of Phil Longo's offenses typically taking 4-6 games to find their strides, and we may be in for an early-season surprise.
Nonetheless, this team will more than likely
this team will lose 4 games. My guess is that they will be: ND, Pitt, @Miami, and
@wake Forest.
What about NCSU?
I've done a deep dive on NCSU and found that only 12 of their 58 2-deep players (
21%) are studs. Given their coach, their relative lack of talent, the revenge factor, and the location of the game, I think UNC is likely to win that matchup.
What's a Good Stud Number?
In the study of 37 teams (ACC and SEC, mostly), the
average team had 30.3% of their two-deep that are studs. The standard deviation is 18.64, so 2/3 of the teams will be in the
range of 11.6% to 48.9%. UNC's 44% is getting there, but still is a long way from the
70s level seen at Alabama. Clemson in 2016 was at
58%. They've bumped up a bit since then, but I haven't gotten to dive into them.
Entertainment Purposes
DON'T GO BETTING YOUR HOUSE on this data. A sample size of 37 is very small and prone to much error. I'll keep gathering this data as I have time, but I thought everyone might like a little perspective on Experience, Recruiting, and Coaching.