Sagarin vs. Vegas, final update

So we've been tracking the Sagarin "Predictor" ratings during bowl season (the same ones that still have OSU ahead of Florida) to see how you would do if you gambled based on them. In each of our fictional wagers you had to bet $110 to win $100. Sagarin says that if you subtract the "Predictor" rating of the underdog from that of the favorite that it suggests what the spread of the game should be.

To refresh your memory each bowl game presented one of 3 scenarios.

• In the first scenario Sagarin and the Vegas casinos disagree on who the favorite is. Under this scenario you would take Sagarin's favorite because you'd be getting points for the team that Sagarin suggests is better. There were only 4 games of the 32 in which this was the case. In these situations Sagarin went 3-1 which translates into $190.
  

• The second scenario is where Sagarin agrees with Vegas on who the favorite is but thinks that the spread should be more than what Vegas thinks it should be. In this case you take the favorite because you would be required to lay down less points than Sagarin thinks the true spread will be. In other words picking the favorite is a value. This was the case in 18 of the Bowl games. In those games Sagarin went 9-9 which translates into -$90.

• The third scenario is where Sagarin agrees with Vegas on who the favorite is but thinks that the spread should be less than what Vegas thinks it should be. In this case you take the underdog because taking the favorite would require you to lay more points than Sagarin suggests the true spread will be. Picking the underdog would be the value. This was the case in 10 games. In those games Sagarin went 6-4 which translates into $160.

So overall you would have wagered on 32 games and gone 18-14 winning a grand total of $260. You would have achieved a return of 7.386% on an "investment" of $3,520 in just 3 weeks. I don't know but it sounds like a lot of money to put at risk for this kind of return.

I wanted to see what happened if we looked at the games grouped by the differences in the Sagarin and Vegas spreads. The theory being that if there really is something to Sagarin's system then we should see better performance when the disagreement between the Sagarin and the Vegas oddsmakers is greater.

• In games wheredifference between the spreads was less than 1 point Sagarin the system went 6-2 for a net win of $380.
• In games where the difference between the Sagarin spread and the Vegas spread was between 1 and 2 points the system went 3-3 with a net loss of $30.
• In games where the difference between the Sagarin spread and the Vegas spread was between 2 and 3 points the system went 5-2 with a net win of $280.
• In games where the difference between the Sagarin spread and the Vegas spread was between 3 and 4 points the system went 3-1 with a net win of $190.
• In games where the difference between the Sagarin spread and the Vegas spread was more than 4 points the system went 1-6 with a net loss of $560.

So when the difference between the spreads was really small the system worked better than when the difference was considerable, which is to say that there's no rhyme or reason to this. I wouldn't gamble any of my money using Sagarin's "Predictor" rankings. Click on the chart below for full size.