Another replay of the model’s piss-poor performance the past two weeks, and all I can say is that teams and players aren’t living up to their early season numbers. So I’ve flipped my thinking a bit. The plan, now, is to dismiss most early season data as irrelevant, and retool the algorithm to focus on the past three weeks of data, starting with the numbers from Week 9 going forward. It’s not as simple as that sounds (relative strength of opposition matters more, and some variable weights will also change), but, for good or ill, I had to do something. (It’s mildly comforting to see that Vegas line-setters aren’t doing much better, with dogs still covering like crazy, and straight-up upsets going at a record pace. All of which is hardly comforting to the bettor, but it is what it is.) Anyway, I did the work, so, fingers crossed, hopefully, Week 12 will turn things around. GLTA
