from Amazon
Tuesday, August 30, 2011
QB aging patterns (and aging pattens in general)
Some interesting stuff from Brian, as he walks you through some of the pitfalls.
Note that the same “last year drop” happens in baseball too. The reason is that this is a sampling bias. If a player has a disaster year, a certain portion of that is bad luck. And a disproportionate number of “last year” performances has more bad luck than good luck. That’s why it looks like you have a big drop in the last year. But that’s only because he wasn’t asked to come back the next year. Well, for those who WERE asked to come back the next year following a disaster year, then, guess what, that disaster year no longer is part of the “last year”. Hence, the bias.
When constructing these aging curves, it is ALWAYS ALWAYS ALWAYS important to pretend that the future has not existed. If you want to know Peyton Manning’s aging pattern from 2011 onwards, then you have to look at other QB who managed to play at a high level at the age of 35. And do NOT look at how (or even if) they played from age 36 onwards in selecting your players. The future is unknown, from the perspective of SELECTING your players.
Once the selection process has taken place, you then do look to their “futures”. Basically, you time travel back to each QB’s point in time such that you are in the same boat as you are with Peyton. And then you travel forward in time for each QB. And since we can’t do that with Peyton without our favorite Delorean, we simply estimate that by looking at all his comparables.
