Started By
Message
Predicting week 6 based off of scoring offense/defense in relation to opponents
Posted on 9/28/14 at 11:23 pm
Posted on 9/28/14 at 11:23 pm
Let me first begin by saying this model doesn't work if a team doesn't have at least 4 games played, so Florida at Tennessee cannot be picked using this. I deliberately did not include the teams with byes here.
A margin of 8 points or less is a tossup. 8-16 is a close game, 17+ is a blowout.
Alabama averages 11 more than their opponents allow and allows 15.5 less than their opponents score.
Auburn averages 14.7 more than their opponents allow and allows 16.9 less than their opponents score.
Georgia averages 13.8 more than their opponents allow and allows 6.9 less than their opponents score.
Kentucky averages 2.2 more than their opponents allow and allows 12.2 less than their opponents score.
LSU averages 15.7 more than their opponents allow and allows 18 less than their opponents score.
Mississippi State averages 17 more than their opponents allow and allows 11 less than their opponents score.
Ole Miss averages 10.3 more than their opponents allow and allows 16.7 less than their opponents score.
South Carolina scores 9.7 more than their opponents allow and allows 5.3 less than their opponents score.
Texas A&M scores 15.6 more than their opponents allow and allows 14 less than their opponents score.
Vanderbilt scored 4.7 less than their opponent allow and allows 1.2 more than their opponents score.
With that, here are the models by team:
Alabama: Ole Miss 24, Alabama 20 (Tossup)
Auburn: Auburn 28, LSU 25 (Tossup)
Georgia: Georgia 49, Vanderbilt 10 (Blowout)
Kentucky: Kentucky 35, South Carolina 21 (Win)
LSU: LSU 32, Auburn 24 (Tossup)
Mississippi State: Tied at 32 (Tossup)
Ole Miss: Alabama 25, Ole Miss 24 (Tossup)
South Carolina: Kentucky 26, South Carolina 25 (Tossup)
Texas A&M: A&M 40, Mississippi State 27 (Win)
Vanderbilt: Georgia 46, Vanderbilt 18 (Blowout)
This predicts a Georgia blowout, a greater than 50% likelihood of a Kentucky win, a greater than 50% likelihood of an A&M win, and two REALLY close games.
A margin of 8 points or less is a tossup. 8-16 is a close game, 17+ is a blowout.
Alabama averages 11 more than their opponents allow and allows 15.5 less than their opponents score.
Auburn averages 14.7 more than their opponents allow and allows 16.9 less than their opponents score.
Georgia averages 13.8 more than their opponents allow and allows 6.9 less than their opponents score.
Kentucky averages 2.2 more than their opponents allow and allows 12.2 less than their opponents score.
LSU averages 15.7 more than their opponents allow and allows 18 less than their opponents score.
Mississippi State averages 17 more than their opponents allow and allows 11 less than their opponents score.
Ole Miss averages 10.3 more than their opponents allow and allows 16.7 less than their opponents score.
South Carolina scores 9.7 more than their opponents allow and allows 5.3 less than their opponents score.
Texas A&M scores 15.6 more than their opponents allow and allows 14 less than their opponents score.
Vanderbilt scored 4.7 less than their opponent allow and allows 1.2 more than their opponents score.
With that, here are the models by team:
Alabama: Ole Miss 24, Alabama 20 (Tossup)
Auburn: Auburn 28, LSU 25 (Tossup)
Georgia: Georgia 49, Vanderbilt 10 (Blowout)
Kentucky: Kentucky 35, South Carolina 21 (Win)
LSU: LSU 32, Auburn 24 (Tossup)
Mississippi State: Tied at 32 (Tossup)
Ole Miss: Alabama 25, Ole Miss 24 (Tossup)
South Carolina: Kentucky 26, South Carolina 25 (Tossup)
Texas A&M: A&M 40, Mississippi State 27 (Win)
Vanderbilt: Georgia 46, Vanderbilt 18 (Blowout)
This predicts a Georgia blowout, a greater than 50% likelihood of a Kentucky win, a greater than 50% likelihood of an A&M win, and two REALLY close games.
8
Posted on 9/28/14 at 11:24 pm to CGSC Lobotomy
quote:
South Carolina: Kentucky 26, South Carolina 25 (Tossup)
I honestly wish I could be rustled by this. But...you might be right.
Posted on 9/28/14 at 11:26 pm to CGSC Lobotomy
The model last week predicted:
A&M model: A&M 49, Arkansas 34
Arkansas model: A&M 40, Arkansas 27
A&M model: A&M 49, Arkansas 34
Arkansas model: A&M 40, Arkansas 27
Posted on 9/28/14 at 11:27 pm to CGSC Lobotomy
This thread would be ALOT cooler if it included Tennessee and FLorida stats.
Posted on 9/28/14 at 11:27 pm to CGSC Lobotomy
What exactly makes you think that margin of victory against FCS and non-Big Five conference teams is a good predictor of performance against SEC competition?
I'm afraid you have wasted your time. Your model is garbage. No offense.
I'm afraid you have wasted your time. Your model is garbage. No offense.
Posted on 9/28/14 at 11:28 pm to Serraneaux
quote:
This thread would be ALOT cooler if it included Tennessee and FLorida stats.
Can't do it because Florida only played 3 games so far.
Posted on 9/28/14 at 11:29 pm to CGSC Lobotomy
quote:
Can't do it because Florida only played 3 games so far.
FAIL
Posted on 9/28/14 at 11:30 pm to TxTiger82
quote:
What exactly makes you think that margin of victory against FCS and non-Big Five conference teams is a good predictor of performance against SEC competition?
It's a predictor of performance relative to opponent.
Using LSU as an example, it held 4 of its five opponents to 15 or more below their scoring averages but only scored 14+ points above their opponents' average points allowed twice (Sam Houston, New Mexico State)
Posted on 9/28/14 at 11:32 pm to Serraneaux
quote:
FAIL
Fine: Florida averages 11.7 more than its opponents allow and allows 4.5 less than its opponents score. Since they've only played 3 games, I can't use them as a predictive model.
Tennessee averages 6 more than its opponents allow and allows 10.9 less than its opponents score.
Using the Tennessee model: Tied at 30 (Tossup)
Posted on 9/28/14 at 11:33 pm to CGSC Lobotomy
quote:
It's a predictor of performance relative to opponent.
But it aggregates information across games when all teams have not played equivalent schedules. You haven't "controlled" for the strength of those opponents.
Worse, your information mostly comes from games against low quality opponents, and you are using it to predict performance against high quality opponents.
I'm sorry man, this is garbage.
Posted on 9/28/14 at 11:33 pm to CGSC Lobotomy
Not bad. Early season is usually the easiest part of schedules, so the variance between points scored and points allowed will the largest in your model.
Posted on 9/28/14 at 11:33 pm to CGSC Lobotomy
Posted on 9/28/14 at 11:34 pm to Serraneaux
So, Tennessee has in essence played a much tougher schedule up to this point and it's a tossup. VOLS by double digits!!!!!
Posted on 9/28/14 at 11:35 pm to AUCE05
quote:
Not bad. Early season is usually the easiest part of schedules, so the variance between points scored and points allowed will the largest in your model.
Also, it's cumulative. The only time this model got a game horribly wrong was A&M at Alabama in 2012.
Predicted Alabama to win by at least 2 scores.
Posted on 9/28/14 at 11:35 pm to Serraneaux
Alabama missed 50-60% of the 4th quarter against FAU, cancelled 4th quarter due to rain, while inside the 10 yard line. Now, it might not do alot to statistics, but it is a noticeable amount of gametime.
This post was edited on 9/28/14 at 11:37 pm
Posted on 9/28/14 at 11:37 pm to TxTiger82
quote:
You haven't "controlled" for the strength of those opponents.
Actually, I have. Case in point for LSU is Wisconsin. As Wisconsin started winning by large margins, it made LSU's numbers better.
Last week, the model had A&M winning on Thanksgiving. This week, it move the game back into the tossup column based off of the performances of all of the teams both schools have collectively played.
Posted on 9/28/14 at 11:38 pm to LoneMDG
quote:
Alabama missed 50-60% of the 4th quarter against FAU, cancelled 4th quarter due to rain, while inside the 10 yard line. Now, it might not do alot to statistics, but it is a noticeable amount of gametime.
and FAU is averaging more than 23 points per game. The shortening of the game preserved the shutout and actually helped Alabama in that regard.
Posted on 9/28/14 at 11:39 pm to CGSC Lobotomy
quote:
As Wisconsin started winning by large margins, it made LSU's numbers better.
That doesn't settle my second point. Wisconsin's numbers improved against weak competition. Then you turn around and use those numbers to predict LSU's performance against Auburn. Doesn't make sense. Why don't you use like-to-like comparisons? Performance against Big Five teams, for instance?
Posted on 9/28/14 at 11:44 pm to TxTiger82
quote:
Performance against Big Five teams, for instance?
Not all Big Five teams are equal. I would rank East Carolina above every nonconference opponent that SEC teams have played except for Wisconsin and Oklahoma.
This post was edited on 9/28/14 at 11:46 pm
Posted on 9/28/14 at 11:45 pm to CGSC Lobotomy
You edited
This post was edited on 9/28/14 at 11:46 pm
Page 1 of 2
Popular
Back to top
Follow SECRant for SEC Football News