Friday, October 17, 2008

Sagarin vs. Vegas, Weekly Picks

Each week, I am pre-selecting 3 games and placing an imaginary wager of $110 (to win $100) on the team that Sagarin's Predictor Ratings suggests.

The system is explained in detail here.

The NCAA football betting lines I'm using come from

So here's this week's picks:

Georgia Tech at Clemson
Vegas Favorite: GT -2.5
Sagarin Predictor Rating for GT: 82.71
Sagarin Predictor Rating for Clemson: 74.39
Sagarin Spread after accounting for home field (2.78 pts): GT -5.54

Sagarin suggests: Take the Ramblin' Wreck of Georgia Tech and lay the 2.5 points.

Miami at Duke
Vegas Favorite: Miami -3
Sagarin Predictor Rating for Miami: 73.82
Sagarin Predictor Rating for Duke: 77.88
Sagarin Spread after accounting for home field (2.78 pts): Duke -6.84

Sagarin suggests: Take the underdog Blue Devils and the 3 points, they are a TD better than the Hurricanes

Vanderbilt at Georgia
Vegas Favorite: Georgia -14.5
Sagarin Predictor Rating for Vandy: 78.03
Sagarin Predictor Rating for Georgia: 86.56
Sagarin Spread after accounting for home field (2.78 pts): Georgia -11.31

Sagarin suggests: Take the Vandy and the 14.5 points.
So we're taking one favorite this week (Georgia Tech) and two underdogs (Duke and Vandy). Good luck.


Jams said...

Just curious: what's the motivation behind the choices of games?

If we're trying to test how useful Sagarin's Predictor Ratings are for placing bets, then the choice of games should certainly not be random. Not applying some system to choose the games (e.g. the games with the biggest difference in lines between Vegas and Sagarin) means that the contest is really just a test of who's better: Vegas or Sagarin.

There's nothing wrong with that, but I'm just curious to know why those specific games are chosen, and how that affects the meaning of results.

Henry Louis Gomez said...


You're absolutely right. The problem for me is that you'd have to look at every game on the slate and calculate the difference between Vegas and Sagarin. With 119 teams it's extremely labor intensive.

When I first did this I did it for the bowl season a couple of years ago and just did all of the bowl games.

When I did it as a regular season feature last year I picked 5 games per week. I generally picked teams I was interested in or interesting match-ups.

Maybe I should request the readers to submit their suggestions for which games to do this for with their reasons.

Truth be told the reason I'm doing this again is that I have a sponsor.

Caffeinated Crocodilian said...

I think a "biggest variance between Sagarin and Vegas" test would be pretty interesting, but you're right, it would be extremely labor intensive... perhaps the top 3 spreads for BCS Schools?

It would be pretty interesting to see if you could set up a 'system' with Sagarin which reliably beat Vegas.