NCAA Football, 2014 week 3. Dave's machine learning algo bets again

Sorry about the ugly image

Yellow highlights are bets where model thinks consensus (right hand column) is "wrong"
Bold are better bets
Bold italic are best.

I am betting $300 on best, $200 on better, and $100 on other good bets.

I had a draft explaining this in more detail that I've been trying to post since Wednesday...
maybe I'll follow up.   We are approaching the sweet spot of college FB inefficiency.  Enough data for me to get an edge but still noisy enough that bettors make some big errors.  Or, perhaps I'm the one making errors.  We'll learn something, anyway.  Frankly, I don't believe some of these.  My model thinks Wake Forest +14.5 vs Utah State is huge but I'm not so sure...  and the way Michigan played last week, well, giving 31 to anyone seems risky.  Not that Miami (OH) is particularly strong.
But, I'll trust the system and put the bets down.  My track record of overriding the model on intuition is really really bad.

The "consensus" is average but Harrah's, Mirage, Wynn & others do not all have the same line - I got a couple of these down with a half point or point better.   This is Tuesday's consensus, by the way, not current.

Have not done much so far.  Left the first week alone and put on 7 last week.  Won big on Pitt but was 2-4 on the others for about break even.

Have a day job and many other hobbies, so might not be doing this everyweek.
Thanks to vegasinsider.com for the chart I swiped with the games in a grid.

I might like to write a little script to upload the data into R and process it there then dump to Excel for presentation formatting...  but don't hold your breath.