So, obviously I didn't get the CC Projections posted. Sorry 'bout that.

I'm running late with the Bowl/Playoff Projections, so it might be tomorrow before I can post them. I've pulled the data, but I just don't have time to roll it up and post it.

(Don't worry - no cheating. What I post tomorrow will be exactly what I would've posted today, if I had time)

Season 128 Bowl and Playoff Projections are now posted. Sorry they're late. Enjoy!

I'm just curious, but why was Southern Connecticut projected to beat my team (Valdosta State)? They did indeed beat me, but your preseason projections showed me a better overall team than them. What's changed since then?

Posted by columbusbuck on 1/21/2017 2:49:00 PM (view original):
I'm just curious, but why was Southern Connecticut projected to beat my team (Valdosta State)? They did indeed beat me, but your preseason projections showed me a better overall team than them. What's changed since then?

My preseason materials are purely talent-based. These postseason projections are almost entirely based on the actual results of the season, inferring ratings for each team and coach through the web of interconnected games played.

There doesn't appear to be much correlation between the spread and win probability. Ratings vs results input like above?

Posted by ebel331 on 1/21/2017 5:33:00 PM (view original):
There doesn't appear to be much correlation between the spread and win probability. Ratings vs results input like above?

There's actually no connection. The spread is determined from one rating model, while the Win Prob is derived from a completely different one.

I know it's awkward, but I do find value in the diversity of modeling.