Sagarin's eigenvector picks

i think i asked once in the past.........

anyone know what this is?.....i knew what an eigenvector and eigenvalue are but that doesn't get me very far.

i thought it might have to do with either 1) focus more on recent results, 2) only use results from this year's game i.e. power ratings don't link to pre-season power ratings (all power ratings systems do link to pre-season ranking, partially because i doubt the 130 teams link up yet..

another clue is that he didn't have eigenvector point spreads for the NFL last week.. there had been no games so couldn't use power ratings based on games so far this year.

BUT....

it spits out crazy and interesting results.

Cal by 19 over ND

BYU by 15 over Oregon

Michigan by 11 over UConn

obviously the michigan prediction is nuts..... BUT, i think they could be interesting lines if you just take them as "Michigan won't come too close to covering vs. Connecticutt".... could be very good for ML betting.

i emailed Ken a few years ago and no response.

thanks in advance :)

Cal has a shot to beat ND. ND lost to an a not-great Marshall team. 

BYU has a shot to beat Oregon handedly. BYU has put up great numbers.

Michigan couldn't cover 50 last week. Can it cover 40 this week? Once a team is up three scores at end of game, it doesn't matter, and they can let the team get 1 score back. So 11 isn't out of the possibility for any two teams, regardless of how good one team is over the other.

Are these three outcomes likely? No.

Are they possible? Absolutely. 

@jimrockford22

 Yooooo found this post and didn’t know if you follow this still. 

fading the eigenvector picks is straight money     It prints all the time. 

@powerade

 I am looking at Sagarin right now and want to make sure that I am understanding correctly. Is he saying that Toledo should be favored by 16.22 points and has a 84% chance of covering the spread. The pick then is to play Akron, take the points. In other words to fade Toledo ? Any insight would be appreciated. Thanks

Interesting