Comments on Advanced NFL Stats: Michael Turner in 2009

Two of the interesting things about regression to ...

2009-08-11T09:48:09.352-04:00

Two of the interesting things about regression to the mean are that it's incredibly trustworthy, on average, but that it's based on unconditional analysis: we're just looking at yards in one season and yards the next.

This has the nice benefit of showing up well on a graph. It has the sadness of making very little use of these incredibly high-powered machines upon which we all type.

I imagine (and for perhaps obvious reasons) that the single best predictor of YPC is...lagged YPC. But as you very helpfully note on your graphs, the R^2 is only ~.2.

I have to guess that you've screwed around with other regressors. Anything noteworthy in terms of raising the R^2? I don't mean overfitting on free parameters, but maybe doing a sample for calibration and a sample for prediction, as you did with your WP function analysis.

Thanks, by the way, these are really interesting.

http://www.pro-football-reference.com/blog/?p=66 ...

2009-07-20T22:17:02.840-04:00

http://www.pro-football-reference.com/blog/?p=66

http://bleacherreport.com/articles/215283-nfl-aging-curves-by-position-rookie-qbs-third-year-wrs-and-age-30-rbs

http://www.feinsports.com/wp-content/uploads/2009/07/aging-curve-rb.bmp

Looks like you're right about decline from the get-go for RBs, at least in terms of YPC.

Brian. Is there an age curve for runningbacks sim...

2009-07-20T21:26:52.417-04:00

Brian. Is there an age curve for runningbacks similar to the age curve for hitters in baseball? In general, batters peak around 27. I'm assuming that RBs peak much earlier, so my quetion may be moot and they may be starting on a downhill path as soon as they turn pro anyways.