Inspired by Fooch's link to snap-weighted average age at football outsiders, I wanted to look at contribution-weighted average age. The first thing I did is see if there was any relationship between snap-weighted average age and team performance. For that, I relied on Football Outsiders DVOA. The plot is below, and you can see that there is no correlation:
Clearly, no pattern. Old teams aren't better. There's a positive correlation (r^2) of .12 on offense, no correlation on defense, and a negative correlation of .12 on special teams, but that is driven primarily by the old, terrible special teams unit of the washington redskins. The 2013 WAS ST unit was one of the worst in history -- throw it out and correlation drops to just .06.
Now that we've established that there is no correlation between age and performance * at the team and unit level, its time to look at the contributions by player and age. At the individual level, we see a very weak correlation between contribution, measured as Approximate Value (AV) in 2013, and years of experience (age minus 22). Age minus 22 is a problem, because players like Kaepernick were 24 year old rookies. But, it is what it is -- I couldn't find experience.
However, what we see here is survivorship bias at work -- old, bad players aren't in the league anymore. So, no relationship. Now we can start to look at age-adjusted contribution. What I did is divide AV by years of experience for each team. the higher the number, the higher the contribution from a younger player. As a result, Peyton Manning had the #1 AV at 19, but an experience-adjusted AV of just 1.1875 -- 419th!
My next step was to sum these up, but I ran into a problem: AV is a function of how good the team is (somewhat), so I had to factor that out, or the better teams would just have a higher youth contribution. so, here is youthAV (y) vs team AV (x):
Pretty clear correlation there. Next step is to factor that out: I created predicted youthAV based on the regression line vs actual youthAV, then dividing actualyouthAV over expectedyouthAV. This gives you a picture of what age groups contributed to the team in 2013 -- it puts aside team performance:
The first thing that stands out is the St. Louis Rams. A huge chunk of team performance came from players who have less experience. compared to the rest of the league, they're a huge outlier -- 3+ standard deviations -- that's a lot. Seattle isn't some team driven by their younger players -- its fairly spread out. The Packers have a lot of young players making big contributions as a percent of the whole.
Now we get to the 49ers. The 49ers are near the bottom of the league when you look at the relative contribution of their younger players. 31% of AV was generated by guys pushing 30 -- 7 or 8 years in the league. However, of those players, Whitner and Brown have left. And Gore's AV will go down this year with fewer carries. Goodwin brought down our number a lot!
But we need to keep getting younger. Teams like the Rams (yikes!) and the Packers are getting their younger players loads of experience right now and that will make them better in the future.