NBA Model Update


I’ve been tweaking the model over the past week, and finally have something that I believe works. Since 12/22, the model is 39-36 (.520). Doing deeper analysis, however, I have made some exclusions that make the model much more accurate:

  • Games with spreads of 3 or less: Any game with a spread under 3 is essentially a toss up, and something to avoid if possible.
  • Games with one or both teams playing back-to-back: Fatigue from playing the night before is something the model cannot account for, so it’s best to just stay away from those games.
  • Games with a model/spread difference more than 7: The model cannot factor injuries into it’s calculations, so games that involve notable injuries often have predicted outcomes much different than the spread. Instead of trying to figure out how to incorporate injuries or which teams have significant injuries, we will just avoid any game in which the model’s outcome is more than 7 points different than the spread.

Using these exclusions, the model is 20-11 (.645) since 12/22 (results here for transparency). Given it’s success, I will start posting picks again. Instead of trying to make a blog post every other day, I will post NBA picks on my Twitter every day. Results will continue to be updated here.


Posted in NBA

One thought on “NBA Model Update

  1. Tweaking mid-season is the surest recipe for disaster. “If only I avoid game types x, y, and z because [insert rationalization], the model kicks ass.” I’ve been down this road too many times myself.

    Only make adjustment during the offseason.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s