September 24, 2022
A couple short updates. First is that the QB model now has its update! So it is based on more years of data.
I don’t stop working! Next will come some improvements to betting lines, but it may not be implemented until week 4. There’s a long list of ideas for little tweaks on all models, so I’m just knocking them off one at a time.
While I’m here, I wanted to point out that this week has an especially shallow variance in outcomes. There are not many, if any, obvious strong choices at D/ST. Expect a lot of variance. I also find it interesting that my spreads model matches vegas lines to a lower extent than I’ve seen except once before. That could go either way, so I’ll be watching with interest. Good luck, and expect some chaos!
September 20, 2022
Data is updated, and everything with the models seems to be working as it should. So there are no urgent updates, but here are a few small things:
As written last week can toggle which default D/ST scores display (“ESPN/Yahoo”) when you visit this website in the future. Just choose your preferred option at the bottom of the webpage! Remember to check your league’s scoring settings.
The kicker model got an update to be fully fitted to 12 years of data, accounting for the difference between low team game scores and high ones. I will try a few more tweaks in the coming weeks, but nothing too drastic.
My backlog is still long. I need to add Yahoo scoring on Patreon, I need to update my articles, I have some ideas for improvements on betting lines (week 2 was good), and I would like to give the QB model the same treatment I just gave to Kicker.
Finally, yes, I would like to get to Accuracy reports as well. I will try to prioritize this for week 3!
Tagged underBackground , Modeling