Yesterday at the Mothership, I took a look at how the number of starters you return can affect your overall momentum as a given season progresses.
Basically, what this says is the following: a team returning fewer than eight starters can probably expect to be about a touchdown worse (in terms of the opponent-adjusted Adj. Points measure) in September than they were last season. They improve by a little over three points in October and November. Meanwhile, teams returning more than 17 starters can expect to improve by between about three and five points the following September, and perhaps somewhat surprisingly, they continue to improve as the year progresses.
The last part is an interesting one. My guess would have been that the most experienced teams come out of the gate strong, then either level out or regress.
As mentioned previously, we already know that "returning starters" is a bit of a flawed metric. If a senior offensive guard started six games, and a sophomore started the other six, is a team returning a starter at that position? And what about a player who was a starter in a previous year but missed last season? We are basically placing 120 teams into one of about 15 categories based on a single, limited number (no team has returned fewer than six starters since 2005, while only four have returned 20 or more), and that isn't a recipe for predictive success. And that's why, when final 2012 projections are released, they will include adjustments for the caliber of starter gone or returning (Did Team A lose six starters, or six first-round draft picks?), the position they play (quarterback appears to carry a bit more heft, though not as much as some might believe), and the level of experience of the players returning. This is a wonderful starting point, however.
I felt this was an interesting concept to observe, but I wanted to take it a little further. The numbers above looked at your total number of starters; how does experience have an impact on offense or defense alone?
(Instead of batching these together, I felt it would be helpful to give you an idea for the sample sizes in these groups.)
Generally speaking, it appears that teams returning fewer than five offensive starters see their production slip a bit (say, three to four points) in September, then slip a bit more as the season progresses. It does appear that "experience," from the 20,000-foot view, continues to play a role as September turns to October turns to November. I would have believed the opposite -- that, as the year goes on, your "new" starters aren't as new, and therefore the playing field evens out a bit. We aren't dealing with enormous swings here by any means, but it does suggest that experienced teams start better, then continue to improve as the year goes on, while inexperienced teams do not.
At first glance, the opposite may be true for defenses.
(Note: for offenses, negative numbers are bad and positive numbers are good. For Adj. Points, as with real points, the opposite is true for defense.)
This was more what I expected to see. Using 2011 TCU as my general anecdote (though, granted, they returned four starters instead of two or three), I figured the path of an inexperienced team from terrible to decent would be a common one. It is much more common on defense than offense. Defenses appear to improve along the way, while offenses do not. This might further the notion that defenses are adapting and improving all season while offenses are not, but we probably shouldn't go that far just yet. I would say we're probably still in the "uncovering data" stage instead of the "reaching conclusions" stage, wouldn't you?
Anyway, using this kind of data, we can certainly take a look (perhaps this weekend or next week) at which teams maybe be best in September or November based on their experience on both sides of the ball.
Final note: I have been using Phil Steele's 2012 Returning Starter data as my early source for experience, but I realized earlier today that the numbers I was using for previous posts included returning special teams starters (kicker, punter). So on the extremes, Boise State only returns five starters, while Tennessee and Texas Tech only return 18 and 19, respectively.