Against my better judgment, I’m starting earlier than usual. The NBA algorithm I run requires a half season of data in order to produce reliable projections, but my nerdy friends ganged up on me, so – here it goes. Specifically, the problem is that the model uses past performance numbers, of teams and players; and, the fewer games played means less data, thus the less reliable are the scoring projections. That said, if it goes bonkers this first week, I’ll pull the plug, and restart in a couple additional weeks. So, a big caveat on the reliability of these projections.
The model uses 77 team and player variables, specific game matchups (including rotations), and an Injury Module that occasionally breaks down. Due to frequent, last minute changes to line-ups, I’ll be posting these projections, most often, as close as possible to the tip-off of the earliest game each day (unless I’m out of town or I have something to do, like today).
GLTA
The model uses 77 team and player variables, specific game matchups (including rotations), and an Injury Module that occasionally breaks down. Due to frequent, last minute changes to line-ups, I’ll be posting these projections, most often, as close as possible to the tip-off of the earliest game each day (unless I’m out of town or I have something to do, like today).
GLTA
