I understand how models work and that taking one game in isolation doesn't prove

Sign up, and you'll be able to ignore users whose posts you don't want to see. Sign up

Report problem with this ad

Jan 3, 2025
2:43:45pm

Schultz is back All-American

I understand how models work and that taking one game in isolation doesn't prove

one way or the other. But for the entire season (700+ games) the model was wrong more than half the time. That's not good. The model is bad.

Report problem with this ad

Schultz is back

Bio page: Schultz is back
Joined: Jan 14, 2013
Last login: Jan 4, 2025
Total posts: 10,092 (961 FO)

Report problem with this ad

Messages

Author

Time

I understand why people hate it this season, but FPI is a decent model

bythenumbers

Jan 3, 2:28pm

as BamaCoug has stated several times today on CB and Twitter FPI was wrong 52%

Schultz is back

Jan 3, 2:31pm

Read my post on model calibration. I don't think I agree with his method.

bythenumbers

Jan 3, 2:35pm

I understand how models work and that taking one game in isolation doesn't prove

Schultz is back

Jan 3, 2:43pm

You shouldn't just evaluate the prediction but the degree of confidence

bythenumbers

Jan 3, 2:51pm

The OP linked the data. Go read it. According to it in 2024, FPI was “correct” 72% of the time, not 48% like you say.

TCuz

Jan 3, 4:07pm

Report problem with this ad

Posting on CougarBoard

In order to post, you will need to either sign up or log in.

Report problem with this ad