Originally Posted by Rain Man
Oh, okay. Nice work.
Maybe then you develop a family of curves, one that includes guys with 10 year careers, another that includes guys with 9 year careers, and so on. I could be mentally deficient, but I still think we have some pollution by the fact that the left side is going to include weaker players who aren't destined for long careers.
The issue with that is that since I only have 5 years of data, if we limit it to guys with 10 year careers, that means we have no data for the first 5 years of anyone's career.
Here's something I came up with. If we limit it to guys who have a full 5 years of data, hopefully we can cover close to the entire range of experience while getting rid of the guys who can't play. Most players wouldn't play for 5 years if they were terrible. I made the graph below by limiting it to guys with 5 years of experience.
The pink line is their average rating, yellow is median, and blue is every player's rating summed, divided by the sum of their snaps, then multiplied by a constant (total snaps of all players divided by total player-seasons). If that makes sense. I'm not sure which line would give the best sense of typical development.