Week 1 vs. Week 15

[RazorHawk]

I would question the logic to weigh a game based on who is favored. Betting lines are only established for wagers and should have nothing to do with any analysis of relative team strength.

I would think the point differential in a game should be considered, but that does encourage running up the score. I do see a difference in Iowa beating Northwestern 21-20 comparing it to Iowa winning by a score of 48-7. If the computers are throwing that away, I think that is a mistake.

I could see programming so that the point differential is a factor, up to a certain point, say 28 or 35 points, thus giving no advantage to beating a team by 70 points.

[Spence]

I would question the logic to weigh a game based on who is favored. Betting lines are only established for wagers and should have nothing to do with any analysis of relative team strength.

I would think the point differential in a game should be considered, but that does encourage running up the score. I do see a difference in Iowa beating Northwestern 21-20 comparing it to Iowa winning by a score of 48-7. If the computers are throwing that away, I think that is a mistake.

I could see programming so that the point differential is a factor, up to a certain point, say 28 or 35 points, thus giving no advantage to beating a team by 70 points.

Which sounds OK, but punishes a defensive league that teams win 17-10, 24-7. Things like that. I think that is the flaw with Sagarin.

[WoVeU]

People are just a little lazy.

A point margin shows something...but you have to bound it the right way. Who has starters in the whole time, is a team more defensive, yet equally as good? There is a pretty simple method. You have to regress to the points of real meaning. If a team won by 21 points or more it is a max win, and treat that however they might. If it is a more defensive game and they won by 13...you should see something like 23-10 but they more than doubled the opponents score, still a max win. To me one game that ended up 55-27 is no more impressive than a 23-10 victory, the treatment should be the same. And when they are writing algorithms there are numerous ways of handling this.