There were a lot of guys in that top-10 that all the models predicted (Cousins, Newton, Watson, Wentz, Brady). There were a few surprises (McCown and Fitzpatrick), but for the most part this week was much closer to expected.
Let’s see how each model did last week:
As I suspected, all the models were better than last week because the world was a little more predictable. The top models are now I, E, D, B, H, and F, with very little daylight between them. I’m glad that my models seem to be doing alright, but a little frustrated that none are standing out. I haven’t tried adding the effect of division, but that didn’t seem like it was statistically significant (https://www.thedatastream.org/data-exploration/2017/9/26/does-in-division-affect-fantasy-scoring). Then again, most of these are statistically weak effects and yet still matter. I may make another model sometime soon.
I like Model I a lot (as the newest model), so I’ll check that one out this week. I’ve been playing Stafford every week, and he hasn’t been doing me any favors. Let’s try changing it up a little.