another model weighs-in - cougarguard.com — unofficial BYU Cougars / LDS sports, football, basketball forum and message board

hyrum · 11-21-2006, 04:46 PM

I guess the Bassett Football Model doesn't have the double-secret PED equation; it has significantly less than 91% chance of winning:

http://members.cox.net/gbassett/foot...6wk13pred.html

jay santos · 11-21-2006, 05:06 PM

Quote:

Originally Posted by hyrum

I guess the Bassett Football Model doesn't have the double-secret PED equation; it has significantly less than 91% chance of winning:

http://members.cox.net/gbassett/foot...6wk13pred.html

Bassett's model is one of many models you can find at http://www.masseyratings.com/cf/compare.htm

He apparently has BYU with a 60-70% chance of winning.

He has BYU #22, Utah #50. The average at mratings is BYU #20, Utah #58.

My model has BYU #19, Utah #54 and a 74% chance to win.

Vegas has a wider MOV than I do, and I'm not sure about their chance to win, but probably greater than my 74%. I'd take Vegas if I had to choose between the different opinions.

Indy's PED Model has BYU #1 and not sure about Utah. Indy would be the first to tell you his model isn't as good a predictor as these other models. It's valuable in the sense that it shows how important pass effiency is.

It is cool that Indy got under the skin of Utes almost as much as Curtis. That's a powerful effect for a dude on the internet.

Indy Coug · 11-21-2006, 05:09 PM

I got under the skin of the Utes? Link?

pelagius · 11-21-2006, 05:12 PM

Quote:

Originally Posted by jay santos

He has BYU #22, Utah #50. The average at mratings is BYU #20, Utah #58.

Just a small note; Gene Bassett's predictor ranking (which is where the 60-70% of winning comes from) actually has BYU as 10th. The #22 comes from his earned rank system.

Indy Coug · 11-21-2006, 05:13 PM

Quote:

Originally Posted by jay santos

Indy would be the first to tell you his model isn't as good a predictor as these other models.

Predictor of what? Odds of winning and losing?

I wouldn't be the first to say that, simply because I've never really looked at other models and their predictive accuracy. Furthermore, I have ignored the first 2/3 of the season as far as PED Differential predictive accuracy goes, simply because I don't believe there is enough statistical credibility until that far into the season, but I haven't really tested that supposition either.

hyrum · 11-21-2006, 05:15 PM

Quote:

Originally Posted by jay santos

Bassett's model is one of many models you can find at http://www.masseyratings.com/cf/compare.htm

He apparently has BYU with a 60-70% chance of winning.

He has BYU #22, Utah #50. The average at mratings is BYU #20, Utah #58.

My model has BYU #19, Utah #54 and a 74% chance to win.

Vegas has a wider MOV than I do, and I'm not sure about their chance to win, but probably greater than my 74%. I'd take Vegas if I had to choose between the different opinions.

Indy's PED Model has BYU #1 and not sure about Utah. Indy would be the first to tell you his model isn't as good a predictor as these other models. It's valuable in the sense that it shows how important pass effiency is.

FWIW, on the page I linked it says BYU #10 , Utah #30. That Massey chart may be out of date.

Suffice to say, the 91% is a bit out in left field.

jay santos · 11-21-2006, 05:15 PM

Quote:

Originally Posted by Indy Coug

I got under the skin of the Utes? Link?

Hyrum's a Ute.

I'm sure they're not thrilled about Harmon's article and scrambling for counter evidence, like Hyrum is.

Indy Coug · 11-21-2006, 05:16 PM

Let me just throw in the comment that the reason why pass efficiency and all of its derivatives are so effective in quantifying performance is because it is a composite of several aspects of a team's play and can be used to quantify both sides of the ball.

jay santos · 11-21-2006, 05:18 PM

Quote:

Originally Posted by Indy Coug

Predictor of what? Odds of winning and losing?

I wouldn't be the first to say that, simply because I've never really looked at other models and their predictive accuracy. Furthermore, I have ignored the first 2/3 of the season as far as PED Differential predictive accuracy goes, simply because I don't believe there is enough statistical credibility until that far into the season, but I haven't really tested that supposition either.

Maybe I'm putting words in your mouth, but I didn't think you really believed BYU was #1 team in nation, Hawaii #3, or Pitt #6. Your model does well to show the importance of PED, but as a model, I wouldn't think you would take it over the models at http://www.masseyratings.com/cf/compare.htm .

Indy Coug · 11-21-2006, 05:20 PM

Quote:

Originally Posted by hyrum

FWIW, on the page I linked it says BYU #10 , Utah #30. That Massey chart may be out of date.

Suffice to say, the 91% is a bit out in left field.

Maybe it is out in left field, but a sample size of 520 games of similar PED Differentials (e.g. ranging from 50 to 75) from a population of nearly 3,300 games should show I'm not completely grasping at straws here either. This includes both home and road games, so the variance in the probability of a road win can't be that much lower than the overall probability.

Additionally, the PED Differential of 70 points between BYU and Utah would place BYU at the top of that grouped bracket, so their odds would actually be slightly higher than 91.7%.

11-21-2006, 04:46 PM	#1
hyrum Senior Member Join Date: Jul 2006 Posts: 860	another model weighs-in I guess the Bassett Football Model doesn't have the double-secret PED equation; it has significantly less than 91% chance of winning: http://members.cox.net/gbassett/foot...6wk13pred.html

11-21-2006, 05:09 PM	#3
Indy Coug Senior Member Join Date: Jan 2006 Location: Between Iraq and a hard place Posts: 7,569	I got under the skin of the Utes? Link?

11-21-2006, 05:16 PM	#8
Indy Coug Senior Member Join Date: Jan 2006 Location: Between Iraq and a hard place Posts: 7,569	Let me just throw in the comment that the reason why pass efficiency and all of its derivatives are so effective in quantifying performance is because it is a composite of several aspects of a team's play and can be used to quantify both sides of the ball.