BYU, Ranking and Lamar

[pelagius]

Originally Posted by jay santos

I have Nevada at 18. I'm pretty sure Colley is W/L only, and he has BYU #26, as well as a high ranking for S Ill. I checked my model, and I have near equal weights to W/L and my modified MOV.

Nevada's a fun one: Sagarin W/L = 8 and Sagarin MOV = 56

[jay santos]

This discussion kind of proves why computer models are more subjective than I'd care to admit.

When I was developing my basketball and football models, I studied other models, and read some analysis on MOV and W/L. I believe MOV is definitely important and should not be ignored. However, I believe some of the arguments against MOV, and empirically I noticed that MOV models seem to favor teams with gawdy offenses and not reward the tough, defensive teams (both football and basketball). You can see the difference this year between UCLA and North Carolina. And in football you saw it between Florida and Ohio State/Michigan.

I ended up using a combination of the two, with pretty equal weighting between W/L and MOV in basketball and slightly more W/L weight in my football model.

[Detroitdad]

How much are MOV models affected by the vagaries of free throw shooting, pulling out the starters for the last few minutes, in basketball, or the cheap touchdowns at the end of the game in football? I think we have all seen games that stretch out or narrow in MOV, and the final scores are not reflective of the actuality of what happened in the game. Is this discrepancy a legitimate concern and is there a way to correct for such effects?

[jay santos]

Originally Posted by Detroitdad

How much are MOV models affected by the vagaries of free throw shooting, pulling out the starters for the last few minutes, in basketball, or the cheap touchdowns at the end of the game in football? I think we have all seen games that stretch out or narrow in MOV, and the final scores are not reflective of the actuality of what happened in the game. Is this discrepancy a legitimate concern and is there a way to correct for such effects?

They are subject to those issues, but the idea would be that everyone is affected equally, and over a season things would average out. Also, I believe most models don't use straight MOV, they use an adjusted MOV, such that beating someone by 10 compared to 20 is more of a difference than beating someone by 20 compared to 30.

[Indy Coug]

Originally Posted by jay santos

This discussion kind of proves why computer models are more subjective than I'd care to admit.

When I was developing my basketball and football models, I studied other models, and read some analysis on MOV and W/L. I believe MOV is definitely important and should not be ignored. However, I believe some of the arguments against MOV, and empirically I noticed that MOV models seem to favor teams with gawdy offenses and not reward the tough, defensive teams (both football and basketball). You can see the difference this year between UCLA and North Carolina. And in football you saw it between Florida and Ohio State/Michigan.

I ended up using a combination of the two, with pretty equal weighting between W/L and MOV in basketball and slightly more W/L weight in my football model.

They don't necessarily have to be that subjective. Find the best predictive equation and go with it.