As I said earlier, that also occurred to me, so I asked Micah about that and he said, first, that he had a separate version of the model that does include some special teams data, but that the predictive benefit of adding it was minor (2+ seasons of sample to add value). Any other tinkering with the model seemed to do more harm than good (I also asked him about factoring in CHIP but he says this just mucked things up), so this is where he landed. It isn't perfect, as should be evident by looking at Minnesota's trend line.
__________________
"The great promise of the Internet was that more information would automatically yield better decisions. The great disappointment is that more information actually yields more possibilities to confirm what you already believed anyway." - Brian Eno
|